A threshold account of last resort in the ethics of war

Abstract

This study aims to demonstrate the essential role that the condition of last resort plays in determining when waging war is exceptionally justified. In recent years, this condition has been criticised despite being considered an indispensable condition for just war. According to the commonly accepted Exhaustion View, the last resort occurs after all other options are exhausted. However, the categorical precedence of non-military measures may exacerbate the situation. In order to address this issue, this study presents an alternative Threshold View of last resort. According to this view, the last resort indicates when a certain threshold is crossed, and the initial presumption against war is overridden by more stringent justifications for waging war. By presenting and defending this view, this study addresses the various objections directed at the last resort condition without undermining the requirement of lastness embedded in it.

Introduction

In the ethics of war, the condition of last resort under jus ad bellum plays a unique role as it defines the temporal dimension of the conflict. According to this condition, every peaceful alternative must be exhausted before a war is authorised. As the saying ‘ultima ratio regum’ (the last argument of kings) goes, every war is regarded as an unavoidable means to resolve international disputes. This is consistent with the relationship specified between Chapters VI and VII of the Charter of the United Nations (UN), which stipulate the peaceful settlement of disputes and taking of measures to address threats to international peace, as well as the relationship between Articles 41 and 42, which provide for non-military and military measures. This study refers to this understanding as the Exhaustion View.

This line of reasoning is, however, contentious: there is no end to measures that fall short of force. How does one determine the point in time when every option – other than military measures – has been fully tested and proven ineffective? Categorically prioritising non-military measures would make it challenging to know when they have been exhausted, leading us to conclude that war must be postponed indefinitely. According to Michael Walzer, ‘Taken literally, [. . .] “last resort” would make war morally impossible. For we can never reach lastness, or we can never know that we have reached it’. ¹

Subsequently, this study presents an alternative Threshold View of last resort, according to which, in waging war, at least one justification must override the presumption against it. War typically causes significant destruction and damage. Consequently, it is reasonable to hold a general presumption against it. However, this presumption is not absolute and has its limit. The condition of last resort indicates the moment when the stringency of the justifications for waging war reaches a specific threshold and outweighs the initial presumption. This account lays out and expands on Walzer’s following view:

But sending troops into battle commonly brings with it so many unanticipated costs that it has come to represent a moral threshold: political leaders must cross this threshold only with great reluctance and trepidation. This is the truth contained in the ‘last resort’ maxim. ²

While it is common to interpret the last resort condition in terms of this threshold metaphor, ³ the literature on the ethics of war has not adequately analysed this threshold’s precise meaning and function. The present study clarifies the meaning of threshold based on the presumption against war rooted in the duty of non-maleficence. Furthermore, it describes how this presumption functions and when it can be overridden using the theory of threshold deontology. By presenting and defending this view, this study attempts to address the criticisms directed at the Exhaustion View by clarifying the meaning of lastness embodied by last resort.

The Threshold View can mount a simultaneous response to several objections raised against the last resort condition. Despite being viewed as an indispensable condition for just war, in recent years this condition has been criticised on several grounds. First, it makes an unreasonable demand on the victims of aggression through the postponement of an armed response. Second, greater harm will descend when war is postponed to the last moment. This study will attempt to test the validity of the Threshold View by demonstrating its response to these challenges.

This study is structured into five parts. First, it interprets the status of the last resort condition under jus ad bellum as a deontological requirement embodied in the presumption against war. Second, it makes a distinction between views that focus on exhaustion and threshold as possible interpretations of this condition. Third, it identifies the threshold indicated in this condition with the presumption against war rooted in the duty of non-maleficence and its general priority. Fourth, it describes a situation when this presumption is overridden by more stringent justifications for waging war by referring to the theory of threshold deontology. Finally, it examines the theoretical persuasiveness of the Threshold View in addressing the objections raised against last resort. Accordingly, this study aims to demonstrate how crucial this condition is in determining when waging war can be exceptionally justified.

I will restrict the scope of this study. Especially in this century, issues of justice in measures not leading to war, or the so-called jus ad vim, have been discussed, with the growing trend of engagement with non-state actors in mind. These issues can impact thinking about the last resort condition. For instance, some have argued that in contemporary situations where the small-scale use of force is envisioned, this condition may become easier to meet, because it does not involve the destructiveness of conventional warfare. ⁴ This study leaves these issues aside and understands war as a military action between states that threatens the territory or sovereignty of another state, such that it constitutes an ‘armed attack’ under international law.

Presumption against war

Before moving on to the Threshold View defended in this study, a premise for last resort, in general, should be presented. ⁵ In my view, the condition of last resort required in the ethics of war rests on the presumption against war. This is a prerequisite condition to be met before resorting to war, along with other just war conditions such as just cause, the prospect of success, necessity and proportionality being considered in the decision-making process. This presumption is also affirmed in an influential Christian document on just war thinking: ‘The church’s teaching on war and peace establishes a strong presumption against war which is binding on all’. ⁶

Considerable controversy exists over whether this presumption should be situated in the just war tradition or other – particularly pacifist – traditions. ⁷ This study’s argument does not depend on these debates, because, as discussed later, I draw this presumption not from the classical texts of just war but from the more widely shared duty of non-maleficence. ⁸ In this section, with some comparison to the existing literature, I show why the condition of last resort is needed in the ethics of war by interpreting it as a deontological requirement embodied in the presumption against war.

The presumption

My interpretation on the condition of last resort begins with the presumption against war. It is based on the principle of non-use of force; even when its use is justified, it is only an exception to this principle. Furthermore, this is a common notion, for it is practised by the international community that states that disputes of any kind should be settled by means other than the use of force. This principle was embraced by the efforts made to outlaw war in the first half of the twentieth century by the League of Nations, the Kellogg-Briand Pact and the UN. The requirement that force should only be used as a last resort is consistent with this principle in the resolution of international conflicts.

The reason why we presumably oppose war in the first place is obvious. It can have harmful consequences, such as the killing and maiming of innocent lives, the destruction of cities and infrastructure, and damage to the environment. People have lost their property and homes. The grim realities of war often include rape and torture on the battlefield, negatively affecting the psychology of those involved in the conflict as well as their relatives. This includes psychological damage inflicted on those who engaged in combat. Consequently, the extent and severity of the harm caused by war provides a strong basis for supporting the presumption against it. ⁹

However, the last resort condition holds that this presumption is not absolute. Lastness implies a limit on the postponement of war. This ensures that war is not initiated unless sufficient conditions are met. Hence, waging war is presumptively unjustified. Even in cases where using force is exceptionally justified, the burden of proof lies with those who favour its use rather than with those who are against it. There is an end to the initial presumption against war, or we can never reach the line of lastness. The decision on whether this line has been crossed depends on certain conditions that could displace the initial presumption.

Against war or injustice?

Some just war theorists have levelled a persistent critique of the presumption against war: the presumption against injustice has shaped the tradition of just war thinking. While this presumption was dominant in classical doctrine during the medieval and modern periods, the presumption against war was introduced relatively recently by pacifists. James Turner Johnson states, ‘the concept of just war does not begin with a “presumption against war” focused on the harm which war may do, but with a presumption against injustice focused on the need for responsible use of force in response to wrongdoing’. ¹⁰

A ‘just’ war is needed to correct injustice – undeniably, this is a central insight that constitutes the foundation of just war thinking, though this ‘just’ is not straightforward. Ultimately (at least until now), war has been an indispensable means of maintaining order and justice in the face of coexistence with other countries. Faced with the urgent need to wage war, one may hesitate, considering the likely consequences; this may result in the loss of sovereignty and independence. However, if war is finally waged, the resulting human and material damage will be widespread in other countries and one’s own country. Thus, war has become a necessary evil to serve a legitimate objective in the anarchical international arena.

In this sense, just war thinking is a typical example of non-ideal theories that, along with the theories of punishment or civil disobedience, offer measures to deal with injustices in the real world. ¹¹ Unlike ideal theories, non-ideal theory does not outline a perfectly just society, seen as a regulative ideal distinct from and in contrast to the existing society, nor does it explain why it is considered so. Since every war causes severe human and material damage at home and abroad, it is – all things being equal – better not to wage war. Like the police and the judiciary, the military is an undesirable but unavoidable option in pursuing justice in a non-ideal world.

Johnson states: ‘To be sure, force is evil when it is employed to attack the justice and peace of a political order oriented toward these goods, but it is precisely to defend against such evil that the use of force may be good’. ¹² The notions of ‘evil’ and ‘good’ here require caution; war is always an evil in that it is an act of harm, and even when it is justified as a way to bring justice, it is only a lesser-evil among multiple choices. Undoubtedly, in today’s international community, war has often remained a political option for the good and bad. However, the good and bad considered here are only a part of the consequences. The presumption against war implies that waging war is different from using knives or cars, which are not inherently harmful but dangerous when used improperly.

Therefore, I do not think that the presumptions against war and injustice are contradictory. On the one hand, war should not be the first resort, because we have presumptive reasons to oppose it, considering the quality and quantity of harm it causes. On the other, war is ultimately acceptable, because correcting injustice in the real world is necessary. Therefore, we must strike a balance between these two presumptions. This is precisely what the Threshold View defended in this study tries to model. Concerns have revolved around the initial presumption against war, perhaps because it can hinder the waging of a necessary and just war. These concerns will be addressed in examining this view later in this paper.

Prudential, consequentialist or deontological?

As in the standard literature, the decision to initiate a particular war should be supported by one or more just causes, made with right intentions and under legitimate authority, and should be a last resort. In addition, it must meet the conditions of the prospect of success, necessity and proportionality in light of the consequences that war brings with it. Among these ad bellum conditions, the last resort condition represents the initial presumption against war. Hence, this condition occupies a unique deontological position within just war requirements that transcends purely consequentialist thinking. ¹³

The status of last resort among various ad bellum conditions has been understood differently. First, last resort is a prudential rather than a deontological criterion. ¹⁴ Classical just war thinking has not always regarded it as the main principle. Just cause and right intention have been positioned as deontological conditions necessary for any war to be justified. However, last resort has been considered a secondary prudential condition, along with the criteria of the prospect of success and proportionality.

This understanding depends on the meaning of the term ‘prudential’. If it refers to a consequentialist criterion as opposed to a deontological one, it is reduced to the following second understanding. Alternatively, if it refers to a practical wisdom that judges carefully according to the circumstances of each case, I do not deny that the judgement of lastness is unavoidably prudential in this sense. It would be naive to assume that any ethical theory, including deontological and consequentialist theories, can provide a generic algorithm for decision-making. I will return to this point later.

Second, the last resort condition is understood as a consequentialist criterion. ¹⁵ Consequentialism is the idea that the rightness of an action depends on the results it produces; an action that produces a better result is desirable and, therefore, correct. This criterion assumes that non-military measures should take precedence over military measures, because military measures are generally presumed to be more harmful and non-military measures less harmful. Thus, last resort has been understood as a corollary of other consequentialist conditions, especially proportionality, which judges the validity of war per se by estimating the goodness or badness of its likely results.

Against this understanding, I assume that last resort is an independent condition in that it departs from the deontological requirement that an act of harm is, whatever the consequences, pro tanto wrong. War cannot be invoked just because it may yield better overall outcomes. In this regard, I agree with Christopher Toner’s point that ‘the moral force of the principle of last resort derives not only from the dire consequences that result from war, but also from a recognition of the moral horror that war is’. ¹⁶ Why and how the ‘moral horror’ that makes war an inherently awful business is deeply connected to the condition of last resort will be returned later in this paper.

No other way but to do it

The literal meaning of last resort alludes to lastness, which draws a line between using multiple strategies to resolve international disputes and is based on sequential order, such as diplomatic negotiations, economic sanctions and, finally, military measures. However, they can also be implemented simultaneously. Based on the initial presumption against war, a decisive line is drawn between military and non-military measures. Specific requirements must be met before the line is crossed; when it is breached, it constitutes the point at which lastness is fulfilled; there is no other way but to do so. Therefore, the situations under which this line is breached must be considered. This section examines the interpretations of the last resort condition provided by the Exhaustion and Threshold views.

Exhaustion view

‘Last resort’ typically refers to an exhaustion of every possibility. ¹⁷ Every peaceful alternative that can lead to conflict resolution must be attempted before force is deployed as a dispute resolution strategy. In an international dispute, using force is permissible only after every possible option to settle it is declared ineffective. Consequently, a lexical order of precedence exists between military and non-military measures; the use of the former is categorically prohibited when there is room to use the latter. I have formulated this interpretation as follows:

Exhaustion View: waging war is only permissible when it is ensured that every viable option other than war has been effectively exhausted.

This formulation is based on several assumptions. First, the judgement on what needs to be done to accomplish an end is inseparable from its prospect of success. A comparison of various means – such as diplomatic negotiations, economic sanctions and military measures – is meaningless unless each one of them can accomplish the same end. Second, it is not necessary to subject all non-military measures to the real test. This formulation is a somewhat lax view of last resort, because it does not strictly necessitate that every conceivable measure – including those that are ineffective – must be actually attempted.

The following objections can be raised against the Exhaustion View. The first problem with this line of reasoning is that there is no end to non-military measures. ¹⁸ For instance, when a diplomatic negotiation fails, it is always possible to proceed to the next round with different conditions. Another example is seen in the use of economic sanctions. Although they may fail to achieve the desired effect, the situation may change when they are imposed over a prolonged period. Under the appeasement policy practised at the Munich Conference in 1938, an appeal to the condition of last resort can be used only as a means to delay final decision, making it nearly impossible for every use of force to meet the definition of a just war.

Second, the last resort condition has been criticised recently because it sometimes invites unnecessary harm. ¹⁹ What if we continue to use non-military measures until they are exhausted, even if they are not the least harmful option? Non-military measures, which should be considered first under the UN’s collective security framework, are generally less harmful than military measures, but not always. For instance, long-term economic sanctions against an illegitimate state may result in greater suffering for its citizens, including women and children, than short-term military sanctions. This categorical preference for non-military measures may worsen the situation.

It may be possible to address these objections to the Exhaustion View by interpreting the criterion of being exhausted in an even more lax manner. Nevertheless, this strategy is self-defeating in the last resort condition. If the criterion is relaxed too much, policymakers will only need to simulate other options quickly in their head or on a calculator and then declare that they have considered the alternatives to a certain extent, rendering the condition meaningless. When interpreted too strictly, the above two objections are invited, whereas when interpreted too laxly, the significance of the last resort condition is diminished. The Exhaustion View does not convincingly explain when the line of lastness is transcended.

Threshold view

To understand the time when a critical point is reached before waging war can be considered permissible, another indicator is needed besides the one that focuses on the prior exhaustion of non-military measures. Consequently, this study proposes the Threshold View, according to which last resort underlines the moment when a specific threshold is crossed after the presumption against war is overridden. Although avoiding war will prevail until this threshold is breached, it can sometimes be outweighed by other more stringent justifications. In this view, the last resort condition can be reformulated as follows:

Threshold View: waging war is only permissible when at least one justification overrides the presumption against war.

The Threshold View is similar to what James Pattison calls the ‘presumptive last resort’. ²⁰ The starting point of this argument is the distinction between doing and allowing. As war causes death and destruction, there are pro tanto reasons to oppose it presumptively. However, this presumption is not absolute. In some cases, it is overridden because it may be justified, all things considered, to cause small-scale harm rather than to allow large-scale harm. Indeed, Pattison also relies on Walzer’s view of the last resort, which this study referred to at the beginning.

Pattison’s argument inspired mine, although they are not identical. First, Pattison includes the additional condition of ‘best means’ that ‘the comparatively best non-military, feasible option(s) should be tried first’. ²¹ This seems close to the Exhaustion View, which differs from the Threshold View to be defended in this study. I do not think it especially important to consider whether non-military measures, in reality or hypothetically, be tried to reach the line of lastness, because we can find another criterion as to whether this line is crossed.

Second, Pattison rejects the threshold deontological approach as a model of the situation in which the presumptive opposition to war is overridden, and proposes a pragmatic approach. ²² This approach is suggested to have some affinity with so-called consequentialising deontology, which incorporates several deontological values that usually emphasise the moral status of an act itself in the overall assessment of the state of affairs. In contrast, I depart from the assumption that the last resort condition is seen as a deontological requirement, and evaluate its constraining nature more stringently. How the threshold deontological approach differs from others and how we can respond to Pattison’s concerns about threshold deontology will be discussed later.

Two types of practical necessity

The Exhaustion and Threshold views share a belief: there may be times when the only choice is to act. However, they differ in their reasons for doing so. In the Exhaustion View, war is employed because every option is exhausted. As per the Threshold View, war is resorted to because the initial presumption against it has reached its limit. To determine whether lastness is reached, a comparison of the stringency of different moral requirements is needed instead of a consideration of the availability of different means of conflict resolution. When overwhelming justifications for resorting to war exist, there is no other choice but to do so.

The difference between the two views may be clarified by understanding the practical necessity concept introduced by Bernard Williams. According to Williams, ‘must’ is a special form of practical necessity judgement, in the sense that ‘Ought is related to must as best is related to only’. ²³ The notion of either the impossibility of other options or the agent’s incapacity is at work here. The ideas of necessity and impossibility correspond to each other. In some cases, impossibility is prioritised. This is because no other options are possible and, consequently, the only possible option is considered necessary. In other cases, necessity assumes priority when, for some reason, a choice is given overwhelming importance, which makes a consideration of other options virtually impossible.

The structural relationship between these two lines of reasoning is presented as follows:

In face of ‘I must’, the other alternatives are no longer alternatives: they become things one cannot do, as, in the other structure, an alternative was something one could not anyway do, and that consideration led to ‘I must’. ²⁴

Bearing this relationship in mind, the Exhaustion and Threshold views conclude that no other course of action can be yielded by different lines of reasoning. The Exhaustion View maintains that the only option – waging war – is necessary because every option has been tested, either in reality or hypothetically, and is found to be virtually impossible. The Threshold View concludes that war is inevitable because the presumption against it has been overridden by other more stringent justifications. While the condition of last resort refers to the lastness of an option, the process of deliberation that leads up to this point may differ.

The differences between the two views may appear minute; in fact, both views may conclude alike in many of their assessment of current international disputes. I have attempted to delineate, as clearly as possible, the different rationales used by both views to determine when the line of lastness is transcended. The specific differences between these views will be discussed again later in this paper. Before proceeding to testing their comparative robustness, the following sections examine the two stages of the Threshold View.

Establishing the threshold

In the previous section, I distinguished the two views on how the initial presumption against war is displaced and reaches lastness. In the following two sections, I focus on the Threshold View and examine how it establishes and crosses the threshold. To set the threshold for waging war, it is necessary to provide an independent reason for preferring non-military measures, regardless of whatever consequences follow from it. In this section, I demonstrate how the presumption against war, rooted in the duty of non-maleficence, constitutes the threshold of last resort.

The duty of non-maleficence

From the outset, some conceptual clarifications are made. A general distinction is made between positive and negative duties toward others. Non-maleficence is understood as a form of omission, which is a deliberate lack of action that does not lead to the occurrence of events. Therefore, the duty of non-maleficence is a type of negative duties for it requires one to refrain from any action, whereas positive duties require one to do so. W. D. Ross listed the pro tanto duties one is usually required to follow when all other things are equal: fidelity; reparation; gratitude; justice; beneficence; self-improvement; and non-maleficence. ²⁵ Among them, only non-maleficence is labelled a negative duty as it is indicated as a negative, not-to-do form.

It is noteworthy that there is a symmetrical relationship between the duties of non-maleficence and beneficence. While non-maleficence refers to avoiding causing harm, beneficence denotes the provision of benefit. Harm and benefit are relative concepts: when an action makes a target worse off than the status quo, they have been harmed; when a target eventually becomes better off than the status quo, they have benefited. In the context of the ethics of war, non-maleficence refers to refraining from acts that would violate the rights to life, limbs, liberty and property of others; beneficence refers to preventing violations of such rights and remedying them. The target of harm and benefit can be an individual as well as a collective entity such as a state or a community.

The priority of non-maleficence

The fulfilment of the duty of non-maleficence does not, however, always align with other duties. The requirement that war should only be waged as a last resort implies a presumptive opposition against it independent of the good or bad outcomes that war brings with it. A more stringent presumption would mean a more pressing need to avoid harm associated with war. Nevertheless, using force may lead to better overall outcomes by saving many lives quickly over a short period. Under such circumstances, should one prioritise the positive duty of lowering the total amount of harm over the negative duty of doing no harm to a permissible extent?

Similar to the adage of ‘primum non nocere’ (first, do no harm) in medical ethics, the duty of non-maleficence is generally considered more stringent, and its fulfilment takes precedence over all duties. According to Ross, ‘the duty of non-maleficence is recognised as a distinct one, and as prima facie more binding’. ²⁶ It also exerts a binding force on other duties. The deontological ethic that an act of harm is pro tanto wrong, regardless of the consequences, is primarily grounded in our intuitions and social norms, exemplified in the distinction between doing and allowing or between killing and letting die.

The different stringencies of duties arise from the contrasting nature of commission and omission. First, omission is less burdensome for the agent, whereas commission is more burdensome. While withholding an action can be observed by anyone, the act of commission requires time, energy and money, and there are natural limits to the resources. For instance, by doing nothing, we simultaneously fulfil our duty not to harm people worldwide. However, we face practical difficulties in fulfilling the good deeds of donating one slice of bread to those in need, given the resources of time, energy and money.

This affects the stringency of the duties when observed. Negative duties that require omission place a lesser burden on the agent and bear a greater stringency in requiring compliance. What is the inconvenience of requiring compliance with the duty of non-maleficence, such as ‘do not kill’ or ‘do not harm’ in our daily lives? The burden that positive duties place on the agent is high; therefore, the severity of the required compliance is not high. We cannot save everyone in need simultaneously. According to the maxim ‘ought implies can’, the scope of positive duties is narrower than that of negative ones. ²⁷

Second, the relationship between omission and its consequences is probabilistic, whereas the relationship between commission and its consequences is causal. A commission directly causes the following result, whereas omission merely allows the result to occur. In order to distinguish whether it is a commission or an omission, the following counterfactual test is used: Would a good or bad event not have happened had the agent not been present? If it would not have occurred, the agent is causally influencing it. If it had occurred anyway, the agent did not contribute to the situation.

This affects the stringency of the duties when violated. As the violation of negative duties requiring omission involves a commission, the agent is causally involved in the result. If we violate the duty of non-maleficence by causing harm, we are causally responsible and thus legally and morally liable for the consequences. Conversely, a breach of positive duties constitutes an omission, and the agent is only probabilistically involved in the consequences. The death of a needy person resulting from our failure to help does not make us guilty of murder.

The prevalence of the priority

The priority of non-maleficence is reflected in the legal system. The high degree of stringency of negative duties to be fulfilled, their observance demanded, and their violation avoided are exemplified by criminal law frameworks, such as the distinction between crimes of commission and omission, as well as different penalties attached to them. Under criminal laws, the pro tanto stringency of the duty of non-maleficence is exemplified by using negative terms aimed to deter crimes, such as ‘do not steal’, ‘do not injure’ and ‘do not kill’, which would subject the culprits to legal sanctions and penalties.

This priority can also be found in debates on the ethical validity of euthanasia. Active euthanasia directly causes the death of a patient, such as by administering a drug injection. In contrast, passive euthanasia allows the patient to die, such as by not keeping the patient on life support. The former is an act of directly violating a negative duty, whereas the latter is an omission that constitutes a violation of a positive duty. In many countries, active euthanasia is prohibited or strictly regulated and can even be punished for murder. In contrast, passive euthanasia is legally permitted in many countries as a method of withdrawing treatment.

In addition, this priority is consistent with a phenomenon known as the ‘omission bias’. This bias reflects our psychological tendency to evaluate the bad outcomes of commissions more negatively than those of omissions. At worst, omission maintains the status quo, whereas commission positively changes it. The negativity resulting from the latter is overestimated by the agent and others beyond what occurs, which is likely to lead to a conservative attitude, suggesting that it is better to do nothing than to do something. Whether this bias is irrational is debatable, but it is a fact that this tendency is widely observed in our society.

Returning to the context of the ethics of war, we maintain a strong presumption against war rooted in the duty of non-maleficence, which generally outweighs the stringency of other duties. ²⁸ An observation of the principle of non-use of force is supported by the negative duty of non-maleficence, even though it may fail to satisfy the positive duty of beneficence. When one is forced to make a choice between the two, the priority of non-maleficence is preferred over meeting the beneficence duty; it is based on the premise that an act of harm should be avoided as far as possible regardless of the consequences. Given the stringency of the presumption against war, there is a pro tanto reason to prioritise non-military measures over military ones. This is because the duty of not doing any harm is generally more stringent than that of lowering the total amount of harm.

Crossing the threshold

The other aspect of the Threshold View is that the threshold is not absolute. The presumption against war begins with the fact that war is a harmful act and is, therefore, wrong. However, it is only a presumption which may be outweighed by other considerations. All forms of non-absolutist deontology have their limits. Like other moral requirements, the duty of non-maleficence is a pro tanto duty that can be overridden by other duties dependent on their relative stringency. Consequently, last resort points to the moment when the initial presumption reaches its limit and the impermissible becomes permissible. ²⁹

Reversed priority

The more stringent justifications for waging war are based on meeting the conditions of just cause, such as defence against aggression or the protection of people from humanitarian crises, as well as other consequentialist considerations, such as the prospect of success and proportionality. When these conditions are met at a certain level of stringency, the presumption against war is inconclusive. The idea that a ‘just’ war is needed to correct injustice describes the conditions under which this presumption can be overridden. The last resort is the temporal point reached when there is a reversal between the requirement to wage war and the importance of avoiding it.

This demonstrates the unique status of last resort among the various ad bellum conditions. According to the Threshold View, the conclusion that there is no choice but to wage war implies that justifications for waging war are overwhelmingly strong. Lastness is seen to have been satisfied when other just war conditions meet specific requirements and finally override the initial presumption against war. The last resort describes when the requirements for waging war and the importance of avoiding it are reversed. The condition of last resort represents an incredibly restrictive aspect of the ethics of war, because it draws its normative force from the presumption against war. ³⁰

A crucial question is at what point the initial presumption against war can be overridden by more stringent justifications. To address this issue, I refer to the theory of threshold deontology. According to this theory, below the threshold, the stringency of the duty of non-maleficence takes precedence over other duties. Despite the stringency of other duties, it does not affect the overall decision-making process. However, when the threshold is crossed, the initial presumption is overridden by the stringency of other duties, such as when the water level rises gradually until it begins to overflow from a dam. Consequently, the irresolvable and often conflicting plurality of pro tanto duties is reconciled by the reversed priority of duties that lies below and above the threshold. ³¹

Threshold deontology has been proposed and evaluated to bridge the gap between deontology and consequentialism. However, several non-consequentialist moral requirements can conflict with the deontological duty of non-maleficence. For instance, the duty of beneficence to save some people, even at the expense of others, can be justified on the non-consequentialist claim for special relations such as associative duties. I understand here threshold deontology as an idea for reconciling the conflicts of multiple duties, along with methods such as balancing and lexical ordering.

Threshold deontology seems to be a pluralist doctrine rather than deontology per se, as it is subjected to different, including non-deontological, moral requirements in the overall decision-making process depending on whether it is below or above the threshold. ³² Besides the question of how to call it, the upshot of threshold deontology is that the threshold remains the same before and after it is crossed; therefore, in some respect, an overall justified act continues to count as a breach of duties. ³³ A pluralist who switches between the types of moral thinking below and above the threshold would not comprehend the moral conflict faced by those who have hesitantly decided to cross the line. To use the metaphor of a dam, the threshold, when it is overridden, continues to exist in the water and to influence its level and flow. ³⁴

Assessing the problem of arbitrariness

There are three main objections to threshold deontology. ³⁵ The first argument is the issue of arbitrariness, because threshold establishment is considered an inherently arbitrary act. The second is the issue of moral ballast, for the harm inflicted below the threshold is merely considered a moral bulwark that does not have any moral import. The third issue is incommensurability, which pertains to a comparison of the good and bad in a utilitarian calculation with the right and wrong of an act itself. Because the second and third objections are concerned with the relationship between deontology and consequentialism, and the subject of this study is the relationship between different pro tanto and partly non-consequentialist duties, I will only refer to the first objection that is pertinent in this context.

Threshold deontology cannot avoid the problem of arbitrariness in establishing a threshold. ³⁶ For instance, can one deliberately take a life to save another person? The duty to avoid killing is usually more stringent than saving others. What happens when there is a need to save five, ten or hundred lives? If the threshold for the number of lives that can be saved is assumed to be between 10 and 100, how and why does one decide on the exact number before the priority of non-maleficence is overridden? Ultimately, it is impossible to reasonably explain the definition of a threshold line everyone accepts, regardless of whether it is indicated by a fixed point or a range.

Consequently, I do not deny that the decision taken to establish the threshold is intuitive and, therefore, arbitrary. When one is confronted with a conflict between the duty of non-maleficence and other duties, there is a lack of definitive response on when the threshold is crossed. Numerous practical factors must be considered, such as the quantitative and qualitative nature of harm and the moral and social attributes of those involved; this is, however, unlikely to be formulated in purely abstract terms. Nevertheless, despite the lack of a consensus on where to draw the limit, all threshold deontologists, except absolutist ones, would agree on whether to draw the limit. Although the choice, such as whether the rule of the road is left- or right-hand traffic, can be arbitrary, this does not mean that it can be denied when it is made legitimately.

The same is true in the context of war. It is challenging to generalise the considerations that must be addressed when a decision needs to be made regarding whether war can be waged as a means to resolve international conflicts. The decision depends on several factors, such as the imminence of rights violations, the political and military nature of the enemy state, the characteristics and intentions of political leaders, the historical relationship between belligerent countries, and the wider state of international relations. In addition, once war has been undertaken, there can be no return to pre-war conditions. In these circumstances, policymakers hesitate to cross the critical line by declaring war. Hence, it is reasonable to think that last resort is not a categorical but a profoundly prudential condition that cannot be forced into abstract propositions.

In this regard, we find another theoretical resource in just war thinking: the condition of legitimate authority, which holds that the declaration of war must be made by those responsible for public order, not private groups or individuals. ³⁷ In a modern system of sovereign states, this primary authority is the sovereign; in a democratic society, it is the people and their trusted political leaders. However, this authority is not entirely free from international law and institutions. For instance, collective security measures are subject to UN Security Council resolutions, and individual or collective rights to self-defence must be reported to the Security Council. In practice, the consensus among these political bodies can be the basis for setting a certain threshold and its validity.

Testing the threshold account (I)

The preceding two sections demonstrated how the presumption against war, rooted in the duty of non-maleficence, constitutes the threshold of last resort. Furthermore, they examined what it meant to cross a certain threshold by using the theory of threshold deontology. In the following two sections, I test the robustness of the Threshold View vis-à-vis the Exhaustion View and demonstrate its responsiveness to two major objections listed in an earlier section, which have been raised about the validity of last resort as a condition among just war criteria.

We begin by addressing the first concern over the condition of last resort: there is no end to non-military measures. Take an obvious example. The international community imposed strict economic sanctions on Russia immediately after its invasion of Ukraine in 2022. Does this mean that a war of self-defence should not be used as a last resort? Should Ukraine and the international community postpone military measures, even for months or years, to wait and see if the economic sanctions would effectively end the conflict? It is unreasonable to request the invaded country to refrain from the use of military means while the invading side is already using force against them. ³⁸

Defensive war and last resort

Just war theorists would not have developed such an argument. ³⁹ The question is where to find the theoretical resources to argue that a war of self-defence can be a last resort for the victim, regardless of the remaining possibility of other non-military alternatives. Dating back to the Gulf War in 1991, requests for the maintenance of additional economic penalties and the deferral of military sanctions were being made before the Allies launched the war in response to the Iraqi invasion of Kuwait. ⁴⁰ Walzer’s view on the last resort, referred to at the beginning of this paper, was raised among the controversies over the use of force during this period. By seriously considering the possibility of other non-military alternatives, the last resort condition could lead to a pacifist call for nonviolent resistance to aggression.

In response, it should be noted that this concern is only applicable to the Exhaustion View and not the Threshold View. In the former, last resort stipulates that non-military measures should take precedence as long as there is a likelihood for them to be effective. In the latter, waging war is permissible not because all non-military measures have been exhausted, but because the justifications for correcting injustice by force are so overwhelming that they override the initial presumption against war. Again, the lastness reached is not determined by the availability of different means of conflict resolution, but through a comparison of the stringency between different moral requirements that argue for or against war.

In a defensive war, then, what are the relevant moral considerations that should be addressed when waging war becomes permissible as a last resort? The first consideration relates to the rights of the defending state. Because the aggression has already occurred, the rights to sovereignty and territorial integrity of the defending side have been violated. The right to self-defence is enshrined as ‘the inherent right’ by the UN Charter, which implies that the defending state may use force, if necessary, to redress a violation of this right. Consequently, there is no inherent reason for them to refrain from doing so. Typically, self-defence assumes a reactive form against a prior unlawful attack that has already occurred, at which point an important condition for forceful defence against the aggressor is fulfilled. This increases the stringency of the justifications for waging war.

The second consideration relates to the liability of the aggressor. When one is liable to some harm, the implication is that one would not be wronged by the harm. When self-defence is employed against aggression, it is an act of harm against a liable target. Although it is presumably wrong to attack the target, it is no longer prohibited in an exceptional situation when the target is not immune to harm. As the aggressor has forfeited their right not to be harmed due to their initial wrongful act, an asymmetrical situation develops in which the defending side is justified in harming the aggressor, but not vice versa. Even when an act of self-defence inflicts damage on the aggressor, it is a situation of ‘you asked for it’. Hence, it is the aggressor – and not the defending side – who should bear the blame, which lowers the stringency of the initial presumption against war.

To use a hypothetical case, suppose a defending state can avoid a full-scale war involving a harsh mainland battle by surrendering some of its territory to an aggressor state. Faced with this option, are they required to obey the duty to retreat? If the demands of the aggressor state are limited but the consequences of ignoring them are catastrophic, the stringency of avoiding force may prevail in terms of the harm from an all-out armed conflict. However, the room for such manoeuvres would be, if anything, limited. Unlike individuals who can retreat without compromising their integrity, states are unable to do so. A retreat in this case is nothing but a partial surrender, and the battle and retreat are no longer comparable options that can serve the same end.

In summary, as soon as aggression is inflicted upon the defending state, it is allowed to employ force as a last resort. Consequently, ‘if an armed attack occurs against a Member of the United Nations’, Article 51 of the UN Charter allows the use of force by a defending state without any prior attempt being made to reach a peaceful settlement of disputes. Indeed, a defending state does not need to apologise or compensate for the damage inflicted on the aggressor state after the war ends. Instead, the aggressor state bears this responsibility. By applying the Threshold View to the last resort described so far, we can identify that moment when waging war becomes legitimate in a manner that is consistent with the exceptional permission given to self-defence by the international community today.

Nevertheless, even when a defensive attack is aimed at a liable target, it does not eliminate the wrongness of the harm. Although an aggressor side may be liable to defensive harm, it does not deserve it. Even though war is not encouraged, it is sometimes justified as a necessary evil. This is why the act of self-defence against aggression is further constrained, in addition to liability, by necessity and proportionality articulated by international customary laws. Regarding the duty of non-maleficence, war, whether below or above the threshold, should still be avoided. Consequently, a case being made to justify self-defence against aggression does not nullify the initial presumption against it.

When to end a just war

The requirement that war should be waged only as a last resort is an important determinant for the continuation as well as termination of it. Because it was initiated with a view to accomplish an end, the justice of war encompasses judgements regarding when it should be ceased. As it is inherently undesirable, war is under moral scrutiny at the point of its termination as well. Its continuation or termination must be constantly re-examined throughout the duration of war. The issues that surround the justice of ending a war have been advanced in recent years as jus ex bello. ⁴¹

What are the relevant considerations in determining the justice of ending a war? The judgements of jus ad bellum and jus ex bello have parallel structures. The former teaches whether a war should begin – when to take up arms – whereas the latter teaches whether it should be stopped – when to lay down arms. It would be misleading to assume that a just war will persist regarding its legitimacy until its end. If this is the case, then the condition of last resort attached to jus ad bellum continues to play a vital role under jus ex bello.

In the existing literature, the condition corresponding to last resort under jus ad bellum has been envisioned based on the Exhaustion View. As a war had already begun, the term ‘last resort’ is inappropriate. Darrel Moellendorf calls the alternative under jus ex bello ‘the pursuit of diplomatic remedies’. It is crucial to see whether a change in the situation can create room for a new diplomatic negotiation that did not exist before. If there are alternative non-military measures that could achieve the initial cause(s) for war, the exhaustion condition fulfilled during the outbreak of the war is no longer satisfied.

I am unsure about the availability of ‘pursuing diplomatic remedies’ as the appropriate point to bring a war to an end. Just as diplomatic negotiations are possible with changes in time and place before resorting to military measures, there is always room to restart new diplomatic negotiations by reframing the terms during the war. The problem of the Exhaustion View arises again. There is ambiguity in deciding when the possibility of peaceful resolutions is granted, depending on whether the condition of pursuing diplomatic remedies is taken strictly or laxly. Instead, I discuss when and how a war should end based on the Threshold View.

The overall decision-making process below and above the threshold is affected by changes in the war situation. At times, belligerents may widen their objectives that transcend their initial justifications to start war. At other times, the conditions for just war that were so overwhelming from the outset may change subsequently, prompting reconsideration of the original decision to resort to war. A limited war may escalate into a full-blown war that is disproportionate to the initial cause that led to its outbreak. The state of belligerency may become more hostile and indiscriminatory toward each other. As the point on the threshold crossed changes, the original decision to wage war can be reversed by the decision to avoid it.

More importantly, the presumption against war does not lose relevance when it is temporarily overridden by the outbreak of war. As noted in the previous section, multiple pro tanto duties remain unchanged below and above the threshold. Despite all justifications, war continues to be inherently undesirable for it violates the duty of non-maleficence. During just war, this presumption continues being weighed against the actual and potential acts of harm, which provides a reason for seeking an early end to war. This is not because of the availability of other options, but because one has every reason to ensure that war can retain its exceptional permission in a volatile war situation. ⁴²

In this sense, decisions to continue or terminate war are fundamentally asymmetrical. On the one hand, the continuation of war is only temporarily supported by the fulfilment of justifications for waging war. However, the termination of war is constantly supported by the presumption against it. Given the circumstances, once a war has been initiated, the default option is not to continue unless an important military objective is at stake. Instead, the default is the non-use of force between the two options of continuation or termination. The decision to continue using force bears the burden of proof of its stringency over time.

By seriously considering the moral, strategic and political importance of the decision to end a war, we come to reconsider the decision to start it. As David Rodin observes, ‘The termination of war is characterised by moral dilemmas that are in the long run morally damaging and prudentially self-harming. Once one realises this, then it becomes clear that the barriers to commencing war must be commensurately strengthened’. ⁴³ The threshold of last resort is placed at the heart of the ethics of war, governing the transition between peacetime and wartime in both directions. The Threshold View provides a consistent basis for evaluating these normative judgements. These possibilities indirectly demonstrate that this view is theoretically and practically more fruitful than the commonly accepted Exhaustion View.

Testing the threshold account (II)

Let us move on to the next objection: compared to the limited use of force, non-military measures do not always constitute the least harmful option. For instance, strict economic sanctions were imposed immediately against Iraq by the UN after it invaded Kuwait in 1990, and were continued long after the Gulf War ended in 1991. Eight years after the Gulf War ended, the United Nations Children’s Fund (UNICEF) estimated that the excess mortality in children under five years of age in Iraq reached half a million. ⁴⁴ The same may happen in other cases. The condition of last resort can sometimes lead to unnecessary harm that could have been avoided had military measures been taken earlier and, therefore, should be jettisoned. ⁴⁵

The harms of economic sanctions

One way to address this concern is to bite the bullet and argue that military measures should be postponed infinitely, regardless of the consequences, unless non-military measures have been exhausted. Despite potentially harmful outcomes, non-military measures take categorical precedence over military measures. The problem with this response is, again, ambiguity in deciding when non-military measures have been exhausted. In the Threshold View, however, there is no need to draw such an extreme conclusion, because the presumption against war rooted in the duty of non-maleficence, which constitutes the threshold of last resort, also opposes certain types of non-military measures.

It should be noted that military measures are not the only means to inflict harm. It can also be caused by intervening in the process of removing harm, if not producing it. ⁴⁶ For instance, denial of access to medications to alleviate the suffering of a person with painful illness would constitute harm to them, because the suffering would not have continued if the agent was absent in the counterfactual test. Harm is inflicted not only directly against a target, but also indirectly by altering its environment. ⁴⁷

Hence, it cannot be denied that economic sanctions can be just as or more harmful as military measures. Certain non-military measures are considered to cause indirect harm when remedial measures are removed from those who have been harmed, which would result in starvation and death. Imposing sanctions to prevent those in need of food and medicines from accessing them can be seen as harming them indirectly. The intentional act of withholding of essential commodities through economic sanctions is considered an act of harm against a target state, a wrongful act that violates the duty of non-maleficence.

That economic sanctions can be no less harmful than military measures does not lower the bar on the latter; the presumption against war would necessitate the use of peaceful measures to resolve conflicts. In the Threshold View, it does not matter whether non-military measures are exhausted, but how far this presumption can retain its initial credibility against the call for harmful response. If the use of both military and non-military measures provided for in Chapter VII of the UN Charter should be avoided because of the direct and indirect harms they may inflict on the target state, the threshold of last resort would then lead us to return to the peaceful settlement of disputes outlined in Chapter VI.

Is doing better than allowing?

Some may dismiss this conclusion by rebutting the priority of non-maleficence per se. Their rebuttal takes the following form: there are innumerable situations where the coercive use of force has helped resolve problems. If one categorically rejects this possibility, they are guilty of committing the same injustice that doing harm entails. While a peaceful resolution is being sought, the current state of human rights violations, food crises, oppression by dictatorships or military developments that threaten the safety of other countries continue to worsen. Those who can do so must respond proactively to such situations, even if it involves enforcing limited force. In summary, compared to the gravity of a positive duty of lowering the total amount of harm, a negative duty of not doing any harm is not as stringent as it looks at the outset.

Two points should be noted when addressing this issue. First, the presumption against war rooted in the duty of non-maleficence is widely shared and deeply embedded in social and institutional settings ranging from national legal systems to international laws. When this priority is abandoned, one must resort to force at any time and place to minimise harm. Harmful actions, including war, are needed in some cases to end the grave atrocities committed at home or abroad; the Threshold View takes this point very seriously. However, one must remember the initial presumption to distinguish between situations that demand action and those that do not, especially given the high uncertainty regarding whether a war can be limited by its initial scope to temporal and spatial limits.

Second, it cannot be known with any degree of certainty that an abandonment of the priority of non-maleficence would lead to an expected justification for waging war. An agent must take more action by positive duties than by negative ones. The United Nations Development Programme (UNDP) had estimated that the ‘1.8 million child deaths each year related to unclean water and poor sanitation dwarf the casualties associated with violent conflict’. ⁴⁸ The lives of half a million of Iraqi children should not have been sacrificed owing to the economic sanctions imposed on Iraq. Such sacrifices continue to be made every year. When the omission of an act is viewed as a serious crime like an act of harm, one would have to do something immediately to save them before using force is considered anyway.

The lastness of waging war rides again

Instead of differentiating between military and non-military measures, one may establish a temporal order based on the premise of least harm to resolve international disputes. ⁴⁹ Under this strategy, last resort requires that the least harmful option be chosen. While non-military measures are generally less harmful than military measures, this is not always the case. Conflict resolution measures should be ranked according to the extent of damage they can inflict on a target. Based on this understanding, ‘an alternative measure does not have to be attempted first if there is no reasonable expectation that it will be less harmful’. ⁵⁰

However, an issue with this strategy is that last resort is subsumed by other just war conditions, particularly by the condition of necessity. The condition of necessity mandates that the belligerents use the least harmful means available to achieve their given end. For instance, diplomatic negotiations, economic sanctions and military measures are compared in terms of the extent of harm that they can inflict. When all can achieve the desired end, the (usually) least harmful option of diplomatic negotiations should be opted, while the harm caused by economic sanctions or military measures is considered unnecessary. Although this assessment is significant, the meaning of lastness embodied by last resort is lost. ⁵¹

In my view, necessity is a modal condition defined in terms of the relative weight of inflicted harms, whereas last resort is a temporal condition that indicates the moment when the choice of inflicting harm becomes exceptionally permissible. When the term ‘necessity’ is used in international legal practice, its implications are ambiguous. For instance, in the 1986 Nicaragua decision of the International Court of Justice (ICJ), one of the leading cases on the use of the right of self-defence, the term seems to be positioned in the temporal sense, assessing the indispensability of using force at the time. ⁵² Incidentally, necessity in the modal sense is sometimes included in the proportionality condition under international law. ⁵³

While the comparison of means in the necessity judgement is based on the scalar notion of the degrees of harm, the comparison of means in the last resort judgement is based on the binary notion delineating the line of lastness. One is required to pursue the least harmful means through a comparison of various military and non-military measures available; this does not contradict the idea that a distinctive loss in value is incurred by the harm inflicted. Therefore, military measures cannot be compared to other measures on an equal moral footing. This may be how last resort has been formulated in some literature as a condition for avoiding a means that is not only harmful but inherently ‘awful’. ⁵⁴

‘War is an awful business. [. . .] However just the cause, we should mourn for all that is lost when war claims its wages from us’, remarked John McCain in a speech in 2004 on the legitimacy of the Iraq War. However, the meaning of his remarks was diluted by the following lastness judgement: ‘there is no avoiding this war. We tried that, and our reluctance cost us dearly’. For many years, the international community has imposed economic sanctions on North Korea and Iran to counter the threat posed to global peace by both countries, but the situation has not improved much. Given the possible argument that more swift and effective measures of any kind can and therefore must be taken, the cases of these countries reveal how crucial it is to seriously view the temporal condition while deciding when waging war can be permissible.

Conclusion

This study is an interpretation of last resort provided by the Exhaustion and Threshold views. Consequently, the Threshold View holds more promise in ascertaining the meaning of lastness embedded in last resort. Furthermore, the duty of non-maleficence is at the heart of the presumption against war. As a pro tanto duty, non-maleficence takes precedence over other duties. Additionally, when the stringency of other just war requirements is elevated, this presumption can be overridden after all things are considered. Finally, the defines of the Threshold View demonstrates its effectiveness in addressing two major objections directed at last resort.

The upshot of this study is that a crucial asymmetry exists in the choice between military and non-military measures, which characterises the lastness of the former. This is consistent with the principle of non-use of force by the international community today. The ethics of war, centred on theories of just war, do not always give utmost importance to the non-ideal, but essential, nature of the choice of waging war. The lastness embodied by last resort is rooted in a strong aversion to war in general and, more broadly, to harm. The ‘great reluctance and trepidation’ that one is compelled to feel when crossing the line of lastness is the truth of last resort, which constitutes an important part of just war thinking as a non-ideal theory. ⁵⁵