1. The State-Centric Ban on Force

Justifications offer reasons for conduct; they clarify that conduct that prima facie violates a rule is lawful. Most discussions of self-defence against non-State actors are discussions about the scope of a justification. They assume that the conduct in question is prima facie, ostensibly, unlawful. More specifically, a State’s self-defence against a non-State actor ostensibly seems to fall foul of the prohibition against the use of force in international relations, enshrined in Article 2(4) of the UN Charter and customary international law. This prohibition is the lynchpin of the contemporary ius ad bellum. It binds all States and is said to extend to State-like entities such as stabilised de facto regimes.Footnote ²⁶ According to the dominant reading, it precludes the use of military force in a general and comprehensive manner, irrespective of its intensity or the motives underlying it,Footnote ²⁷ and it does so as ius cogens, with peremptory force.Footnote ²⁸

While general and comprehensive, the ban on force is purposefully limited: it prohibits the use of force by States ‘in their international relations’. This was traditionally read to refer to ‘the international relations between States’,Footnote ²⁹ notably the non-consented use of force on the territory (or within other recognised spheres of influence) of another State – as opposed to action ‘within its own boundaries’.Footnote ³⁰ It is today commonly extended to uses of force across internationally accepted armistice lines or against stabilised de facto regimes.Footnote ³¹ That is a modest extension, though, and the cornerstone rule against force has rightly been described as ‘state-centric’.Footnote ³² This in turn affects the subsequent analysis in two crucial respects.

2. Wheat and Chaff

To begin with, the focus on the prohibition against force helps separate the wheat from the chaff, viz. distinguish military responses that require justification from those that do not. The starting point is straightforward: military responses require justification if – and only if – they ostensibly violate the prohibition against force. This is the case for uses of force that affect ‘the international relations’ in the State-centric sense described above. Military strikes targeting non-State actors based in another State are the obvious example in point. By contrast, other military responses do not need to be justified under the ius ad bellum. The most significant exclusion concerns military responses ‘at home’: in civil wars, against domestic terrorists, etc. Perhaps surprisingly (but indisputably), the prohibition against force, that ‘cornerstone of the human effort to promote peace in a world torn by strife’,Footnote ³³ ignores the most significant forms of contemporary conflict.Footnote ³⁴

The other exclusion is more discrete; it illustrates that different non-State actors cannot always be treated alike. True enough, in a world carved up by States, a prohibition that precludes the use of force on the territory of another State covers most military strikes against non-State actors operating from outside the responding State’s own boundaries. Much of the scholarship, especially on counter-terrorist responses, takes this for granted and seems to treat exceptions as oddities.Footnote ³⁵ However, it is worth noting that one group of non-State actors has traditionally operated from outside State jurisdiction: namely pirates, who operate on the high seas or in a place outside the jurisdiction of any State.Footnote ³⁶ Military action against pirates on the high seas is anything but an oddity. However, it is subject to a special set of rules: these impose restrictions,Footnote ³⁷ but do not implicate general ius contra bellum provisions.Footnote ³⁸ For reasons of convenience, they are left to a side here – but the brief reference suggests that the ius ad bellum does not necessarily treat different categories of non-State actors alike.

3. Asymmetry

Other than helping separate wheat from chaff, the focus on the ban on force draws attention to a complicating factor – in fact the major complicating factor that muddies discussions of self-defence against non-State actors and that can be described as a problem of ‘asymmetry’.

Asymmetry is a consequence of the particular focus of the prohibition against force. State-centric as it is,Footnote ³⁹ that prohibition addresses non-State actors only in an indirect manner. It requires a sequence of events involving one (attacking) non-State actor and one (responding) State to be appreciated on the basis of rules devised to apply between States. In that inter-State perspective, non-State actors hardly feature.Footnote ⁴⁰ They do not feature as attackers, as their initial attacks as such do not implicate the ius ad bellum.Footnote ⁴¹ And they do not feature as targets in their own right either: the responding State’s military reaction violates the ban on force only if it is directed against targets on the territory (or in other spheres of influence) of another State.Footnote ⁴² If it does, it needs to be justified. And while this typically will be the case, the obligation remains State-centred: what needs to be justified is a breach of international law vis-à-vis another State. Military responses against non-State actors therefore cannot meaningfully be thought of as a bilateral relationship between responding State and targeted non-State actors.

Surprisingly often, this straightforward point is not made. Kimberley Trapp makes it when noting that any ‘true exception to the prohibition on the use of force … must in some way excuse the violation of the host state’s territorial integrity’.Footnote ⁴³ In fact, as part of the State-centric ius ad bellum, this is all a ‘true exception’ needs to do.

On balance, State-centrism probably increases the scope for force to be used lawfully against non-State actors: host State consent in particular has become a convenient remedy, relied upon to ensure the legality of crossborder strikes. Yet absent such consent, responding States are ‘caught’ by the problem of asymmetry. What is motivated as a response against a non-State attack has to be justified within a set of rules operating between States. Between the logic of the response (State/non-State) and the rationale of the applicable legal rules (State/State), there is no equivalence: the two are asymmetrical. Much of what follows is an attempt to come to terms with this asymmetry.

1. Self-Defence as the Justification of Choice

The analysis so far establishes that military responses against armed attacks by non-State actors usually require justification. So why should they be justified as self-defence? The short answer to this question is ‘because States choose to rely on self-defence rather than on alternative justifications (other than host State consent and a Security Council authorisation)’.Footnote ⁴⁴

Self-defence emerged as the justification of choice for perfectly plausible reasons, but not over night. Nineteenth- and twentieth-century debates were characterised by much greater diversity. In the pre-Charter era, depending on type of attack and the timing and locus of the response, doctrines such as necessity,Footnote ⁴⁵ armed reprisals and hot pursuitFootnote ⁴⁶ were popular – at times in conjunction with self-defence. The adoption of the Charter, with its stricter system of prohibition and exceptions, initially did not affect this. The different explanations, recast as justifications for prima facie breaches of the ban on force, continued to be relied on; in fact, some doctrines such as ‘hot pursuit on land’ were only fully developed in the Charter era.Footnote ⁴⁷

Over time, though, the consolidation of the Charter regime led to a more streamlined discourse, in which self-defence became central.Footnote ⁴⁸ This it became largely by default, as alternative justifications fell out of favour. The fate of necessity illustrates the process. During the twentieth century, necessity has come to be recognised as a circumstance precluding wrongfulnessFootnote ⁴⁹ and is now very much en vogue in other areas of international law.Footnote ⁵⁰ Within the ius ad bellum, while significant as a limit on self-defence, it is no longer pleaded as a self-standing entitlement justifying recourse to force.Footnote ⁵¹

Claims based on hot pursuit have suffered a similar fate. Primarily espoused by South Africa and Southern Rhodesia during the 1970s and 1980s, they were always received coolly.Footnote ⁵² After the independence of Zimbabwe, and the end of South Africa’s incursions into ‘frontline States’, they are no longer advanced.Footnote ⁵³ Armed reprisals do not fare much better: declared illegal in prominent legal texts,Footnote ⁵⁴ the concept has effectively been abandoned as a justification for the use of force.Footnote ⁵⁵

With alternative justifications losing ground, self-defence today seems to be the ‘last claim standing’. Since the 1990s, it has become absolutely dominant: as Lubell notes, it is indeed the ‘sole avenue for legitimizing unilateral forcible action by states against non-state actors in the territory of other states’.Footnote ⁵⁶

2. Reasons and Implications

Why self-defence should have become the justification of choice is not difficult to see. Unlike necessity, hot pursuit, and armed reprisals, the right to self-defence is expressly recognised as an exception to the ban on force, which is strong enough to justify conduct that ostensibly violates a peremptory norm.Footnote ⁵⁷ Pursuant to Article 51 of the Charter and customary international law, the ban on force does not ‘impair the inherent right of … self-defence if an armed attack occurs’. A State invoking self-defence can be criticised for overstretching an exception; but it acts from within the accepted system. States relying on it of course need to argue that self-defence is available against non-State actors. However, unlike States invoking necessity or hot pursuit, they do not have to make the broader claim that the contemporary ius ad bellum recognised exceptions not explicitly mentioned in the Charter. Reliance on self-defence to justify military responses against non-State actors thus is the ‘safer option’.

Safety comes at a price though. The conscious choice of States to treat military responses against non-State actors as a question of self-defence means that claims are to be presented within the parameters of that particular exception. This has one obvious implication and a further one that is not so obvious. The obvious implication is that military responses against non-State actors have to meet the conditions and modalities of self-defence. Four of them stand out:

1. – The first concerns the triggering event: justified as self-defence, military responses can only be directed against attacks that qualify as ‘armed attacks’: the dominant view restricts this to qualified uses of force, or even the ‘most grave forms of the use of force’.Footnote ⁵⁸

2. – The second concerns the timing of the response: particular care must be taken if action in self-defence is meant to avert future attacks; the right, after all, is recognised only ‘if an armed attack occurs’.Footnote ⁵⁹

3. – The third concerns the scope of the right: reliance on self-defence implies acceptance of the twin conditions of necessity and proportionality – each of them offering some flexibility, but both in principle ‘well settled’.Footnote ⁶⁰

4. – The fourth, and final, condition goes to the purpose of the response: a military response presented as self-defence needs to serve a defensive purpose. This functional element is by no means easy to apply in practice, but it can help to exclude action that is retaliatory or punitive.Footnote ⁶¹

All this, one might say, comes with the terrain. It is the price States have to pay for relying on the ‘safer option’ of self-defence. In reality, things are more complicated. The four limitations cannot be seamlessly applied to military responses against non-State actors. But that is a question of operationalising a set of conditions – which is beyond the scope of the present study.

3. More on Asymmetry

1. The Continuing Appeal of Custom

Before pursuing arguments about asymmetrical self-defence, it is necessary to address a curious methodological problem that besets the debate. As indicated in the Introduction, there is considerable uncertainty about the proper source of self-defence, which is variably located in Article 51 of the UN Charter, custom or a blend of both.Footnote ⁷⁴ Whether the different methodological choices actually affect outcomes is difficult to say with certainty;Footnote ⁷⁵ but they make it more likely that debates turn into a dialogue of the deaf.

The source of the right of self-defence, to be sure, was a matter of dispute in the early days of the UN era. The wording of Article 51 of the Charter invited discussion: pursuant to its first sentence,

As this is clearly a rather roundabout way of recognising an exception to the ban on force, it may not have been far-fetched to argue, in the early years of the United Nations, that a broad (customary) right of self-defence of the pre-Charter era would continue to exist (‘un-impaired’, as it were) alongside Article 51 – and that such a broader, customary right did not depend on a prior ‘armed attack’.Footnote ⁷⁶ But from early on, proponents of such a reading struggled to explain how their ‘black hole’ approachFootnote ⁷⁷ could be squared with the Charter’s desire to impose a strict set of rules regulating recourse to force.Footnote ⁷⁸

The debate about self-defence against non-State actors raises issues of a different character. It turns on the understanding of the term ‘armed attack’ (a term used in Article 51), not on the possibility of self-defence against other acts: in this sense, too, it is a debate that can be had ‘within the system’.Footnote ⁷⁹ And yet, uncertainties about the proper source of law persist; and if anything, the appeal of custom has increased over time. To illustrate by reference to prominent contributions: Daniel Bethlehem’s widely discussed set of principles on self-defence is described by their author as an attempt ‘to work with the grain of the UN Charter as well as customary international law, in which resides the inherent right of self-defense’.Footnote ⁸⁰ Olivier Corten retraces ‘Controversies Over the Customary Prohibition on the Use of Force’.Footnote ⁸¹ Others seek to have it both ways: hence Tom Ruys’ book on ‘Armed Attack’ and Article 51 of the UN Charter evaluates ‘Evolutions in Customary International Law and Practice’.Footnote ⁸² According to André de Hoogh, ‘most authors set forth their analysis or make arguments with a view to establishing the content of a rule of customary international law.’Footnote ⁸³

2. The Proper Focus: Treaty Law

The prominence of custom in discussion of self-defence against non-State actors is puzzling. To be sure, it would be perfectly understandable to locate the debate in custom if Article 51 of the UN Charter was merely a renvoi to extra-Charter law. But that view is not seriously maintained today. Self-defence is generally held to be ‘regulated by both customary and conventional norms’;Footnote ⁸⁴ these two rules – even where ‘they appear identical in content’ – ‘retain a separate existence’.Footnote ⁸⁵ Of the two separate rules, the conventional one, Article 51, does not address all issues expressly; it needs to be applied in conjunction with concepts such as necessity and proportionality that it does not mention.Footnote ⁸⁶ But its terms (‘armed attack’, ‘occurs’, etc.) do provide guidance on some issues, and they do so as treaty law.

The prominence of custom in debates about self-defence against non-State actors would also be understandable if the customary right of self-defence against armed attacks (however construed) could justify violations of the ban on force irrespective of their source. But such a view is difficult to square with general principles governing the interaction of treaty and custom. More specifically, it faces three significant objections. First, it presupposes a highly unusual view of co-existing rules of treaty and custom. Such co-existence, to be sure, is anything but exceptional. International law in many fields contains formally separate, but substantively similar, rules.Footnote ⁸⁷ But such overlap is regularly dealt with by according primacy to treaty law rules. Custom (in the terms used by Dinstein and Thirlway, respectively) usually ‘remains invisible’: it ‘is eclipsed’ and ‘reced[es] behind the treaty’, waiting to ‘reappear in its full vitality … whenever the treaty no longer blocks it from sight’Footnote ⁸⁸ – e.g. when a treaty, on grounds of jurisdiction, cannot be applied (as in the Nicaragua case). Yet as debates about self-defence against non-State actors are not so jurisdictionally limited, one should expect customary self-defence to ‘recede’ behind the Charter rule.

Second, on a more practical level, it is by no means sure how the customary rule on self-defence should preserve its autonomy from the Charter rule. With the United Nations nearing universal membership, international practice can no longer easily be allocated to one particular source. Again, this problem is not specific to self-defence: custom and multilateral treaty norms, while retaining their separate identity, are becoming increasingly amalgamated in many fields. The doctrine of sources does not preclude this, but increasingly accepts ‘Entangled Treaty and Custom’.Footnote ⁸⁹ Treaty law, as will be discussed in section IV, can evolve; and treaty interpretation is to take account, in various ways, of subsequent practice in the application of the respective treaty.Footnote ⁹⁰ Where the treaty in question, like the UN Charter, is widely ratified, the lines between ‘subsequent treaty practice’ (for the purposes of treaty interpretation) and ‘international practice’ (for the purposes of ascertaining custom) become blurred.Footnote ⁹¹ The acceptance of State conduct within the UN, and of UN resolutions themselves, as factors relevant to the ascertainment of custom reinforce this trend. In short, a customary right of self-defence (assuming it were not eclipsed) could hardly be insulated from developments in treaty law.

Third, and most significantly, even if it existed autonomously from its UN Charter equivalent, it is very difficult to see how a customary right of self-defence could provide an effective justification for military responses on foreign State territory.Footnote ⁹² Whatever the position under customary international law, it bears reminding that States using force against non-State actors operating in another State prima facie violate international law’s prohibition against force, which is recognised in treaty law and in custom. If customary self-defence is to be an effective justification for such conduct, it would need to justify both violations. The customary rule on self-defence would not only need to apply alongside its treaty law equivalent, but it would need to ‘trump’ the Charter system. It is one thing to argue customary rules should continue to exist, and to apply, alongside treaty-law rules governing the same ground. It is quite another to claim they should disapply the treaty law. This would turn the regular application of the lex specialis principle on its head.Footnote ⁹³

These three objections do not strictly rule out the possibility of an autonomous right of self-defence, recognised under customary international law, which would justify uses of force that are ostensibly illegal under custom and treaty. But they suggest it is a remote possibility, and one that – given the blurring of lines between customary law and treaty – would be very difficult to establish. Few contributors addressing questions of self-defence against non-State actors as a question of customary international law bother to engage with the three objections. Their preference for custom (as opposed to treaty law) may reflect a desire to focus on practice (undoubtedly a central element of custom, whose crucial role in treaty interpretation is not always appreciated) or an unwillingness to take the terms, context and telos of the Charter rules seriously. It may also simply show a laissez faire approach to the sources of international law, perhaps verifying Kammerhofer’s charge of ‘methodological weakness’.Footnote ⁹⁴ Either way, it remains puzzling – and it ignores the main thrust of the preceding discussion: that self-defence, as regulated in Article 51, raises questions of treaty law. It is addressed as such in the following section.

3. Consequence: A Question of Treaty Interpretation

The focus on self-defence as a treaty-based right structures the inquiry. Whether States can invoke self-defence in response to armed attacks by non-State actors is a question of treaty interpretation, to be addressed on the basis of the principles codified in Articles 31 to 33 of the Vienna Convention on the Law of Treaties (VCLT), which reflect custom. In their customary guise, these apply to treaties predating the VCLT (such as the Charter).Footnote ⁹⁵ They apply to a ‘constituent instrument of an international organization’, but are ‘without prejudice to any relevant rules of the organization’.Footnote ⁹⁶

These considerations provide the framework for the subsequent analysis. This framework (unlike an inquiry based on custom) is capable of taking seriously the written treaty text, as read in light of its context and the Charter’s object and purpose.Footnote ⁹⁷ At the same time, it is flexible enough to accommodate the subsequent conduct of States, which can influence the interpretation of Article 51, either as an authentic or supplementary means of interpretation, as ‘[w]ords are given meaning by deeds.’Footnote ⁹⁸

Before approaching Article 51 on the basis of this flexible framework, two preliminary points need to be addressed. The first concerns the normative environment of Article 51, which is part of a treaty establishing an international organisation with an institutional structure. It is widely accepted that the interpretation needs to reflect this fact. This is typically achieved by broadening the range of actors whose conduct affects the interpretative process. More specifically, the ‘deeds’ that matter for the purposes of treaty interpretation are not only those of States; the subsequent practice relevant for the construction of self-defence encompasses the conduct of UN organs alongside that of member States.Footnote ⁹⁹ In fact, to the extent that secondary acts of the UN are designed to concretise the meaning of Charter provisions, they are ‘privileged sites’ of treaty interpretation.Footnote ¹⁰⁰

If UN practice can, in principle, be easily integrated into the regular framework of treaty interpretation, the impact of judicial decisions poses a greater challenge. International courts – notably the ICJ – have pronounced on different aspects of the ius ad bellum.Footnote ¹⁰¹ Read properly, these decisions have limited binding force, but can carry considerable persuasive power.Footnote ¹⁰² Binding precedents they are certainly not: this is contradicted by Article 59 of the ICJ Statute and equivalent provisions.Footnote ¹⁰³ Nevertheless, a judicial decision will often carry significant authority as a ‘means for the determination of rules of law’ as per Article 38(1)(d) of the ICJ Statute. In this respect, decisions of the ICJ have been described as ‘persuasive precedents’Footnote ¹⁰⁴ and as ‘beacons, guides and orientation points’.Footnote ¹⁰⁵ Those descriptions seem apt, as they highlight the authority of a judicial decision, but also clarify that such authority is not a given: a decision needs to ‘persuade’, and whether it does so has to be assessed. The judicial voice, then, is an important one, but the Court ‘does not have the last word’Footnote ¹⁰⁶ in questions of treaty interpretation.

The second preliminary consideration concerns the peremptory status of the ban on force. This status, in the view of some commentators, affects the principles of interpretation.Footnote ¹⁰⁷ Mary Ellen O’Connell makes the point emphatically; according to her, ‘[g]iven the nature of peremptory prohibitions, … valid interpretation must not result in a weaker norm. Logically, peremptory norms may expand, not contract. … [D]iluting and contracting the prohibition on the use of force through interpretation is impermissible.’Footnote ¹⁰⁸ This approach is informed by a desire to construe the ban against force effectively. However, it fails to appreciate that such effective interpretation ought to proceed from the general regime of interpretation codified in Articles 31 to 33 of the VCLT. As a plea for a special regime of interpretation, the approach faces two obstacles. For one, it seems to assume that the ‘true meaning’ of peremptory norms were certain and timeless.Footnote ¹⁰⁹ But that is difficult to sustain. To illustrate by reference to matters not at issue here, the meaning of ‘force’ in Article 2(4), or of the ‘special intent’ requirement of the prohibition against genocide, is not God-given; it needs to be established – and in this process, the means of interpretation mentioned in Articles 31 to 33 of the VCLT have their place.Footnote ¹¹⁰

But the plea for a special regime faces a more fundamental obstacle. It misconstrues the relationship between the ban on force, and the recognised exception of self-defence. ‘[B]uilt into the very nature of the UN system’,Footnote ¹¹¹ self-defence operates on the same hierarchical level as the ban on force.Footnote ¹¹² Arguments about the peremptory status should reflect as much: what is peremptory is the rule against unlawful uses of force. Action in self-defence is simply not unlawful. Arguments about asymmetrical self-defence are advanced from ‘within the system’ of the UN Charter.Footnote ¹¹³ The rule is limited by the exception, which operates on the same level.

None of this is to criticise the quest for an effective construction of the ban against force. Nor should it be read as a plea for a laissez-faire approach to treaty interpretation. However, it suggests that if peremptory norms were subject to a special regime of interpretation, then this would have to apply to the ban on force as limited by self-defence. For the crucial question relevant here – is self-defence available against armed attack by non-State actors? – ius cogens offers fairly little.

***

On the basis of these clarifications, the threshold question can be approached in its proper normative setting: it requires the interpretation of a treaty clause, Article 51, that enshrines the right of self-defence against armed attacks as an exception to a State-centric ban on force. According to the International Law Commission’s (ILC) much-quoted formula, this interpretation is ‘a single combined operation’, whereby different means of interpretation are ‘thrown into the crucible’, to allow for their ‘interaction’,Footnote ¹¹⁴ which accords each means of interpretation ‘appropriate emphasis’.Footnote ¹¹⁵ For reasons of convenience, the subsequent discussion presents arguments relevant to this ‘single, combined’ process in two steps. The bulk of the analysis (section IV) looks at the approach of States and UN organs. Treaty interpretation, however, is more than a tracing of practice. It proceeds from the text (‘armed attack’), presumed to reflect the parties’ understanding of the proper scope of self-defence. Put differently, words are not only given meaning by deeds; they also have meaning as words, an ordinary meaning, which context, object and purpose help elucidate. A surprisingly large number of writers on self-defence fail to explore literal, contextual and teleological arguments, give short shrift to them, or use them uncritically.Footnote ¹¹⁶ Section III illustrates what they are missing: it takes Article 51 seriously as a rule of treaty law.

1. The Immediate Context: The Terms of Article 51

A first contextual argument flows from the very phrase that contains the terms ‘armed attack’.Footnote ¹³² As noted above (section III.A), Article 51 says nothing on the identity of the attacker, but specifies that it has to be directed against a ‘Member of the United Nations’, i.e. a State.Footnote ¹³³ This is indicative; it makes it difficult to argue that Article 51, as a general matter, simply presupposes some form of State nexus. The argument can be put in the form of two questions. If such a nexus were assumed – why would the victim of an attack be expressly qualified as a UN member (State)? Conversely, if the provision expressly stipulates that only States can exercise self-defence under the Charter, does this not suggest that an armed attack, not so qualified, could emanate from a broader range of actors? This is not a necessary inference. But it is worth noting the contrast between the two sides of Article 51: the victim is expressly described as a State, the attacker not so, even though a qualifying term could easily have been added. This suggests that the silence in the text of Article 51 may have meaning.

2. Arguments Derived from Article 2(4) of the Charter: Conventional Wisdom

The most popular contextual argument suggesting that the silence should be ignored – and that self-defence is only available against a State attack – draws on the relationship between Articles 2(4) and 51.Footnote ¹³⁴ Aspects of that relationship have been addressed already; the two most relevant ones for present purposes are that, first, Article 51 is an exception to the rule found in Article 2(4) and, second, that Article 2(4) in the main precludes the use of force directed against other States.Footnote ¹³⁵ These starting points indeed favour a ‘State-oriented’ interpretation of the right of self-defence that construes the notion of ‘armed attack’ like the ban on force.Footnote ¹³⁶ As an exception justifying military conduct, self-defence can easily be construed as a ‘responsive’ right permitting symmetrical reactions against uses of force that violate Article 2(4). If it is the flip-side of the coin, then it should matter that the front side formulates a State-centric prohibition.Footnote ¹³⁷

Water-tight, or logically necessary, this conclusion is certainly not: perhaps self-defence is more than the flip-side of the ban. But it has considerable appeal, at least as long as the focus remains on unilateral uses of force.Footnote ¹³⁸ Appeal, because prohibition and exception would follow the same (State-centric) rationale. Appeal, because it would reflect the close link between Articles 2(4) and 51, which are regularly presented as a ‘package’, the latter necessary to make the former palatable. Appeal, finally, because self-defence would operate in a straightforward manner: the exceptional response would only be available against States committing a qualified wrong.Footnote ¹³⁹ The symmetric, State-centric construction of self-defence would, to put it bluntly, make eminent sense and fit smoothly within a Charter regime shaped by the State-centric ban on force.

3. Questioning the Conventional Wisdom

All of this, to be sure, is ‘assailable’,Footnote ¹⁴⁰ and has been assailed. Three contextual counter-arguments could sever the link between prohibition and exception. Two of these can be dealt with relatively briefly; the third requires a fuller analysis.

The first counter-argument is an argument from the contrary. It, too, looks at the interplay of self-defence and the ban on force, but it emphasises differences. That Article 2(4) is State-centric only throws – so the argument runs – the openness of Article 51 into starker relief. The difference in wording, argues Andreas Zimmermann, ‘seems to imply, by way of an argumentum e contrario, that Article 51 … does not require any inter-state situation’.Footnote ¹⁴¹ But this only goes so far. Every argument by analogy can in theory be turned into an argumentum e contrario. The question is whether the ratio legis warrants a ‘contrarian’ construction.Footnote ¹⁴² The close link between Articles 2(4) and 51 would rather seem to support a concordant reading.Footnote ¹⁴³

The second counter-argument questions the idea of symmetry between Articles 2(4) and 51. If these two provisions were truly symmetrical, every breach of Article 2(4) should trigger the right of self-defence. However, that is not the case: according to the dominant view, there is a ‘gap’ between the two, as self-defence is available only against qualified uses of force.Footnote ¹⁴⁴ Perhaps, then, other forms of asymmetry should not be ruled out? But that, too, only goes so far. The accepted case of asymmetry between Articles 2(4) and 51 concerns the intensity of force, not the actors. As the two asymmetries are different, the move from the accepted (intensity) to the disputed one (actors) requires a leap of faith.

4. In Particular: Chapter VII of the Charter

More powerful is a third contextual counter-argument, which is rarely made.Footnote ¹⁴⁵ It derives from a comparison between Article 51 on the one hand and Articles 39 and 42 on the other. Articles 39 and 42 – like Articles 2(4) and 51 – form part of the Charter’s ius ad bellum. As read today, they permit the use of force by States with a Security Council mandate.Footnote ¹⁴⁶ While that institutional setting is particular, from the perspective of the responding State, Articles 39 and 42 operate in much the same way as Article 51: they justify conduct that ostensibly violates the ban on force.Footnote ¹⁴⁷ What is more, Article 51 reinforces the link between unilateral and institutionalised reactions by precluding self-defence when the UN’s collective security mechanism is activated. All this suggests that Articles 39 and 42 can inform the contextual interpretation of the ‘armed attack’ requirement.

Article 39 clarifies that by adopting collective security measures. The Security Council can respond to acts of aggression, breaches of the peace or threats to the peace. Just as Article 51 is worded openly, so is Article 39: nothing in the text suggests that Security Council responses are dependent on some form of unlawful inter-State conduct, let alone the use of force.Footnote ¹⁴⁸ Yet for some time, such an inter-State, symmetrical reading enjoyed considerable support. Article 39 was read to presuppose an unlawful use (or threat) of force in the international relations between States.Footnote ¹⁴⁹ Lacking a foothold in the wording of Article 39, this interpretation was supported by contextual arguments, notably the close link with Article 2(4).Footnote ¹⁵⁰

Twenty-five years after Desert Storm, such readings are but a remote echo from a distant past. It is beyond doubt today that the Security Council may authorise military measures in situations not involving an inter-State force.Footnote ¹⁵¹ In fact, the Council has hardly ever responded to breaches of Article 2(4).Footnote ¹⁵² Perhaps more importantly, it has also authorised military measures against non-State actors.Footnote ¹⁵³ On the face of it, States implementing such mandates would have violated the prohibition against the use of force. Yet covered by Articles 39 and 42, their conduct did not violate Article 2(4) and had to be endured by the host State.

This development is part of a new, robust construction of Chapter VII of the Charter that accommodates what Kimberley Trapp calls ‘actor pluralism’.Footnote ¹⁵⁴ And while some of the Council’s other arrogations of competence are viewed sceptically, its decision to overcome an inter-State construction of Chapter VII has met with general approval. For the debate on self-defence, this development is instructive in two respects. First, it shows that inter-State readings of another openly worded Charter provision (Article 39) at some point enjoyed support – and were abandoned subsequently. Perhaps the law of self-defence could have undergone a similar transformation. And, second, a quick glance at the development of Chapter VII undermines the claim that, under the Charter’s ius ad bellum regime, military force could only ever be lawfully used in response to prior breaches of Article 2(4):Footnote ¹⁵⁵ uses of force authorised under Chapter VII can clearly be asymmetrical.Footnote ¹⁵⁶

***

Contextual arguments point in different directions. The immediate context of the term ‘armed attack’ supports a broad construction of Article 51: the provision qualifies the victim of an armed attack (which must be a State) but not the attacker. By contrast, the close relationship between Articles 51 and 2(4) seems to support a State-centric construction. However, the popular argument derived from Article 2(4) rests on a narrow comparison that ignores the Charter’s other exception to the ban on force, viz. force authorised by the Security Council. Once Articles 39 and 42 are appreciated (which they hardly ever are), the argument derived from Article 2(4) loses much of its force. The Charter’s ius ad bellum scheme, if looked at as a whole, is significantly more diverse than commentators exploring the relationship between Articles 2(4) and 51 recognise. This suggests that, contrary to a commonly held view, contextual arguments do not support a State-centric construction of self-defence. On balance, they would seem to offer support for a reading that accepts the possibility of asymmetrical self-defence.

1. Ensuring the Maintenance of Peace and Security

The main teleological argument supporting a State-centric construction proceeds from what has been referred to as the UN’s ‘purpose of all purposes’,Footnote ¹⁶⁰ the maintenance of international peace and security. Under the Charter scheme – so the argument runs – the UN is to maintain peace and security primarily through measures of collective security, while unilateral military action is strictly limited by Article 2(4). Reinforced by the preamble’s emphatic statement against the ‘scourge of war’, this purposive reading is said to mandate a narrow reading of entitlements to use military force unilaterally: ‘as a product of the horrors of the Second World War inspired by the desire to end the “scourge of war”, the whole object of the Charter was precisely to limit the scope for unilateral use of force as much as possible and to subject it to the control of the Security Council.’Footnote ¹⁶¹ In the present context, unilateral force in self-defence is most effectively limited if the trigger event is narrowly construed; and this is best achieved by limiting it to (armed) State attacks.Footnote ¹⁶²

If all of this sounds a bit too good to be true, then that is because it is. More specifically, the purposive argument just summarised suffers from two problems. First, it proceeds from a traditional understanding of the Charter’s peace and security scheme, which emphasises the absence of military conflict between States. This traditional understanding remains prominent, but is no longer dominant.Footnote ¹⁶³ A brief passage from the UN Secretary-General’s Report In Larger Freedom outlines a twenty-first century view of the institution’s ‘purpose of purposes’:

The move from one (inter-State) to another (broader) understanding of peace and security is a gradual process. But that it is well under way seems difficult to dispute.Footnote ¹⁶⁵ And this is sufficient to weaken arguments premised on an equation of peace and security on the one hand and ‘international war and conflict’ on the other: for in the new, ‘multidimensional’,Footnote ¹⁶⁶ understanding, non-State actors are certainly capable of threatening international peace and security. Action against them could well be portrayed as an attempt to maintain peace.

The point need not be explored, as the purposive reading faces a second, more serious, problem:Footnote ¹⁶⁷ it is premised on a selective view of the Charter’s peace and security design. No doubt the two aspects mentioned by its proponents, viz. collective security and the ban on force, are central. However, so is the right of self-defence, which the purposive argument just sketched out conveniently ignores: an ‘un-impaired’ part of the Charter regime, self-defence operates as a limitation on the general ban on force, and is itself limited by collective security action. Read properly, the Charter’s object is to maintain peace and security by banning military force, but only to the extent that force is not used in self-defence. In other words, the maintenance of peace and security as the Charter’s primary purpose is operationalised through particular Charter provisions, of which Article 51 is one. These provisions concretise the Charter’s object of maintaining peace and security; and a solution to problems of interpretation ought to be sought through them, not by reference to the abstract notion of peace and security.Footnote ¹⁶⁸ Adapting the terms used by the ICJ in response to a broad, purposive construction of a jurisdictional treaty, one could say that ‘[a]lthough … States had expressed in general terms in the [Charter’s Chapter I] their desire to [maintain peace and security], their consent thereto had only been given in the terms laid down in [the specific provisions operationalising that purpose].’Footnote ¹⁶⁹ Once this is accepted, the purposive argument collapses: it depends on a selective analysis that simply bypasses Article 51.

2. Facilitating Effective Responses

Whereas purposive arguments favouring a State-centric construction may appear lofty, considerations in support of an asymmetrical understanding of self-defence seem a little prosaic. They are quite rare. Some commentators suggest that Article 51 sought to permit effective responses against real threats or real attacks – which is taken to favour a broad construction of the armed attack requirement.Footnote ¹⁷⁰ Similarly, others claim that the Charter’s peace and security design looked to substantive factors (such as the intensity of an attack) and should not depend on the status of the attacker.Footnote ¹⁷¹ But neither prong of the argument is really convincing. The purpose of self-defence (as opposed to that of the Charter) is not as such a relevant teleological consideration. Moreover, the Charter is incredibly ‘status-conscious’ in some fields (including in formulating a State-centric ban on force). And self-defence could still be effective if it applied only to State attacks.

***

All things considered, it may be understandable that commentators accord ‘relatively little weight’Footnote ¹⁷² to teleological considerations. These simply do not offer compelling arguments either way. None of this should come as a real surprise. The Charter seeks to integrate competing goals into one overarching framework. While the UN’s main purposes may ‘have proved timeless and universal’,Footnote ¹⁷³ the Charter regime has been adapted and can no longer be reduced to ‘one single, undiluted object and purpose’.Footnote ¹⁷⁴ It is perhaps no wonder teleological considerations lack focus.

1. Sleepwalking into an Inter-State Reading

Looked at from a contemporary perspective, the most significant feature of the debates during the first decades of the UN era is that the inter-State framework was accepted without serious discussion. There was no equivalent, in the early years of the United Nations, to today’s protracted debates between ‘restrictivists’ and ‘expansionists’.Footnote ²⁰⁸ States (and UN organs) seemed simply to take for granted that an armed attack had to be one with some level of State involvement: they sleepwalked into an inter-State construction. As a result, straightforward statements explicitly rejecting the possibility of self-defence against non-State actors are rare.Footnote ²⁰⁹ The contemporary evidence supporting an inter-State construction is essentially indirect. It is a by-product of debates about the level of State involvement required to turn an attack into an ‘armed attack’. As will be discussed more fully below,Footnote ²¹⁰ this level remained disputed for a while. Crucially, though, even States that took a broad view of self-defence argued on the basis of an inter-State framework.

Greece’s complaint to the United Nations, one of the earliest invocations of self-defence in the Charter era, in many ways set the tone. The debates of 1946/7 were prompted by ‘private attacks’, but centred on the role of sponsoring States: Yugoslavia and Albania had allegedly supported armed bands or tolerated their activities – and thereby had ‘breached the peace’ and committed an ‘aggression’ against Greece.Footnote ²¹¹ From 1946 until around the late 1980s, in debates about attacks by non-State actors and possible responses thereto – from Burma/China (1953)Footnote ²¹² to the French raids on Tunisia in the late 1950s and the manifold Israeli strikes against Palestinian targets in Jordan, Lebanon and elsewhereFootnote ²¹³ – the focus was squarely on the respective host States. The problem of armed attacks by non-State actors was treated, in Brownlie’s words, as one of ‘State complicity in, or toleration of, the activities of armed bands directed against other States’.Footnote ²¹⁴ It was a problem of determining the required degree of State involvement: that there had to be some State nexus was generally assumed – asymmetrical self-defence was to be approached via the ‘Westphalian avenue’.Footnote ²¹⁵

2. The State Nexus: Initial Flexibility

If the inter-State framework was taken for granted, it was anything but rigid, but initially understood flexibly. To some extent, this flexibility resulted from the fact that States did not proceed from any preconceived general regime of attribution defining for which acts a State would have to bear responsibility; quite to the contrary, the matter was treated as a question of the (primary) rules governing recourse to force.Footnote ²¹⁶ Different views as to the required State nexus came to the fore during the long-standing debates about the proper legal qualification of instances of ‘indirect aggression’, notably held within the UN General Assembly.

Indirect aggression was recognised as a pressing problem in the 1950s and 1960s. In a number of consensus resolutions, UN member States seemed ready to equate indirect and direct forms of force (and thereby to dilute the State-centric character of the prohibition).Footnote ²¹⁷ The Friendly Relations Declaration, adopted in 1970, consolidated the process: it expressly qualified a broad range of ancillary acts as self-standing violations of the ban on force, among them a State’s ‘acquiesc[ence] in organized activities within its territory directed towards the commission of [acts of civil strife or terrorist acts]’.Footnote ²¹⁸

Quite how this should affect the understanding of the right of self-defence proved controversial. Should a State’s ‘participation in the use of force by unofficial bands’,Footnote ²¹⁹ or perhaps even its acquiescence, qualify as an armed attack? The practice of the 1950s and 1960s, reflected in the Friendly Relations Declaration, left the matter open. For a significant period of time, the inter-State framework was applied flexibly. In the words of Olivier Corten: ‘In the first two decades after the Charter’s entry into force, no decisive precedent can … be invoked [for or against the permissibility of self-defence against acts of indirect aggression].’Footnote ²²⁰

3. GA Res. 3314 and Beyond: A Stratified General Framework

In the course of the 1970s and 1980s, the initially flexible framework came under pressure. While concrete disputes about self-defence continued to be handled pragmatically,Footnote ²²¹ the general framework was stratified, and the required State nexus construed restrictively. This was a gradual development, but with the benefit of hindsight, one can identify three important catalysts: a) the General Assembly’s Definition of Aggression; b) the interpretation given to that resolution in the ICJ’s Nicaragua judgment; and c) the agreement, in the course of the ILC’s work on State responsibility, on a set of narrow rules of attribution.

1. Significant Practice

To begin with practice, a significant number of States – during the entire period of the Cold War – responded to ‘private’ armed attacks by using force on the territory of another State, or clearly asserted a right to do so.Footnote ²⁵² They did so even though the other State concerned was neither responsible for sending the irregular bands nor ‘substantially involved’ in their activities in the sense of Article 3(g), narrowly construed. In line with the dominant understanding, responding States regularly justified their conduct from within an inter-State framework, but construed this framework flexibly. In the typical setting, responding States asserted that host States had supported armed attacks, or at least offered armed groups a safe haven. In some instances, host States were also said to bear responsibility for attacks they had failed to suppress; in still others, ‘support’ and ‘harbouring’ rationales were combined. Whatever the details, a more tenuous State nexus was considered sufficient to trigger a right to respond.

France’s raids into Tunisia during the Algerian War of Independence are illustrative.Footnote ²⁵³ They were considered necessary to ‘assurer [France’s] légitime défense’Footnote ²⁵⁴ against FLN attacks from Tunisia: Tunisia, argued France, had to endure them, as it had permitted the FLN to use its territory and failed to suppress its activities.Footnote ²⁵⁵ France’s assessment echoed Greece’s above-mentioned claimsFootnote ²⁵⁶ as well as views expressed by India during the early stages of the Kashmir conflict.Footnote ²⁵⁷ In India’s view, Pakistan’s support for armed bands crossing into Jammu and Kashmir amounted to an ‘act of aggression against India’; in response, India claimed to be ‘entitled, under international law, to send [its] armed forces across Pakistan territory’.Footnote ²⁵⁸

From the late 1950s onwards, variations of this broader understanding of self-defence would be embraced by a number of countries embroiled in national liberation conflicts. Portugal (in the 1960s and 1970s),Footnote ²⁵⁹ South Africa (during the 1970s and 1980s, once ‘hot pursuit’ claims had lost appeal)Footnote ²⁶⁰ and Rhodesia (during the 1970s)Footnote ²⁶¹ invoked self-defence to justify military action against neighbouring States that provided support and sanctuary to armed groups.Footnote ²⁶²

Lesser known is Morocco’s reliance on essentially the same argument in the West Sahara conflict during the late 1970s.Footnote ²⁶³ Following repeated attacks by Polisario forces operating from within Algeria, Morocco claimed a right to ‘poursuivr[e] ses agresseurs sur et hors son territoire’, i.e. into Algeria, which was accused of having armed, trained, financed and sheltered Polisario fighters.Footnote ²⁶⁴ From the early 1980s, Turkey began to advance similar claims.Footnote ²⁶⁵ In 1983/4, foreshadowing subsequent operations on a much larger scale,Footnote ²⁶⁶ it mounted armed incursions against PKK bases in northern Iraq, which was accused of serving as a sanctuary.

While Turkey’s justification remained elusive, other countries offered greater specificity. Israel did so frequently. From 1948, it regularly invoked self-defence to justify its military actions against Palestinian fedayeen based in Arab countries.Footnote ²⁶⁷ Having initially pointed to the close operational ties, or the willing cooperation, between Palestinian fighters and host States,Footnote ²⁶⁸ from ‘the 1970s, Israel gradually adopted a broader version of the “harbouring” rationale’ pursuant to which self-defence could be exercised against a host State that ‘was either unwilling or unable to prevent cross-border attacks from taking place’.Footnote ²⁶⁹ In the 1980s, the United States moved towards the same position. The so-called ‘Shultz doctrine’ asserted a right to respond forcibly against armed attacks by terrorists and guerrillas, and to do so by using force against foreign countries that ‘support, train, or harbour [them]’.Footnote ²⁷⁰

All this suggests that decades before 9/11, the argument for asymmetrical self-defence had been clearly articulated by a considerable number of States, and acted upon with some regularity. In the debates of the day, the problem of asymmetry was typically not addressed ‘head on’, but from within an inter-State framework in which some State nexus was sufficient. As regards the character of this nexus, much of the practice – echoing views put during the preparation of GA Res. 3314 – concerned instances in which a host State had allegedly actively supported armed groups. However, the lines between support, the provision of safe havens, and mere toleration were blurred, and at least Israel began to claim that self-defence could be used against States unable to suppress armed groups.

2. Inconsistent Responses

Unsurprisingly, these assertions of a broadly construed right of self-defence prompted debate: express responses were rarely supportive, and often hostile. And yet, a careful analysis reveals a nuanced picture. Broad claims of self-defence were by no means consistently rejected, and when they were, disputes about the scope of Article 51 played a limited role.

To begin with the latter aspect, the conduct of South Africa, Rhodesia, Portugal and Israel met with widespread criticism, often expressed in strongly worded UN resolutions.Footnote ²⁷¹ To give just two examples, the Security Council in 1985 ‘condemn[ed] vigorously’ Israel’s raid on the PLO Headquarters, which it qualified as an ‘act of armed aggression … in flagrant violation of the Charter of the United Nations’;Footnote ²⁷² three years earlier, it had ‘[s]trongly condemn[ed] the apartheid regime of South Africa for its premeditated aggressive act against the Kingdom of Lesotho’.Footnote ²⁷³ Below the surface, things were not quite as clear. The debates about these and other incidents indicate that the conduct of States such as South Africa, Southern Rhodesia, Portugal and Israel ‘was condemned on many different grounds’.Footnote ²⁷⁴ Facts and evidence often were crucial. (Had the responding States been able to make out their allegation of State support for armed bands? Had there been cross-border raids at all?)Footnote ²⁷⁵ As to the law, many military actions were considered to be disproportionate, or punitive and thus outside the scope of self-defence.Footnote ²⁷⁶ Most importantly, overarching perspectives were hardly conducive to a positive assessment.Footnote ²⁷⁷ During the 1960s and 1970s, broad understandings of self-defence were prominently espoused by States defending deeply unpopular causes (and losing ones at that), viz. colonialism, apartheid and military occupation. With friends like these, asymmetrical self-defence really needed no enemies. In fact, claims of South Africa, Southern Rhodesia, Portugal and also Israel were typically rejected a limine: as Christine Gray observes, they ‘were undermined by the fact that the states invoking self-defence were regarded as being in illegal occupation of the territory they were purporting to defend’.Footnote ²⁷⁸ ‘The right of self-defence could not be invoked to perpetuate colonialism and to flout the right of self-determination and independence’Footnote ²⁷⁹ – statements like these were common, and overshadowed finer normative arguments about the required degree of State involvement. In fact, so dominant were these ‘different grounds’Footnote ²⁸⁰ that the minutes of Security Council debates yield ‘virtually no statements directly dealing with the applicability of Article 51 to cross-border attacks by non-State actors’.Footnote ²⁸¹ To reiterate: none of this suggests the broader view of self-defence adopted by Israel, South Africa and others was widely endorsed. But it was hardly ever rejected specifically.

The international response to claims of asymmetrical self-defence by States other than Israel, South Africa, Portugal and Southern Rhodesia points in the same direction. Outside the colonial or apartheid context, broad claims to self-defence attracted far less opprobrium. India’s assertion of a right to ‘send … armed forces across Pakistan territory’Footnote ²⁸² in response to that country’s support of cross-border raids was not rejected.Footnote ²⁸³ When the Security Council engaged with Morocco’s similar claim in 1979, a number of States mentioned Polisario’s struggle for national liberation, but Morocco’s view of self-defence went unchallenged.Footnote ²⁸⁴ The legality of Turkey’s incursions into northern Iraq in the early 1980s was hardly discussed; and even France’s cross-border ripostes into Tunisia, which did target a national liberation movement, were not firmly censured.Footnote ²⁸⁵

3. Circumstantial Evidence and Interim Assessment

A careful scrutiny of international practice thus suggests that the international community seemed prepared to live with the occasional instance of self-defence even where it did not meet the criteria of Article 3(g), narrowly construed. The international response often seems to have turned on questions of facts and evidence, or on overarching views of the rightfulness of a particular struggle. Some host State involvement in an armed attack seemed indispensable, but even during the 1970s and 1980s, well-founded claims of support (or even harbouring) could be sufficient to expose a host State to responses on its territory.

When looking beyond disputes in which a right to self-defence was exercised or clearly asserted, the picture becomes fuzzier still. Attacks by armed bands were often described as acts of aggression or armed attacks, irrespective of whether they could be traced back to a foreign State. The Security Council on occasion condemned as ‘aggression’ or ‘armed attack’ the conduct of mercenaries on foreign soil.Footnote ²⁸⁶ In his detailed analysis, Claus Kress identifies a significant number of further disputes, in which States considered private armed attacks within a framework of self-defence, or referred to them as ‘aggression’ in a legal sense. His analysis includes protests by Guatemala against cross-border attacks by rebel forces,Footnote ²⁸⁷ for which (according to Guatemala) Honduras and Nicaragua bore responsibility.Footnote ²⁸⁸ It also comprises a curious dispute, in which two States – Portugal and Zaire – agreed that support for mercenaries could amount to an armed attack.Footnote ²⁸⁹

The value of this evidence varies, and some of it may not amount to much.Footnote ²⁹⁰ States typically responded within their boundaries (so that the ius ad bellum was not implicatedFootnote ²⁹¹), or stopped short of expressly claiming a right of self-defence. However, the statements form part of the broader normative discourse about the scope of the ius ad bellum. Perhaps they are best qualified as ‘circumstantial evidence’: of lesser weight, but a further indication that terms such as ‘aggression’ and ‘armed attack’ could be construed broadly.

Viewed in this perspective, the circumstantial evidence supports the modest argument advanced here: that in its treatment of particular disputes, the international community could be prepared to accept a tenuous State nexus. In a relevant number of instances, States advanced broad constructions of self-defence. Their views never remotely reflected the views of ‘all UN members’ and hence could never ‘establish the agreement of the parties regarding the interpretation of [Article 51]’.Footnote ²⁹² But, of course, the practice of Israel, South Africa, Turkey, France etc. could be relevant as ‘other subsequent practice’ in the sense of the ILC’s Draft Conclusions. As such, it could help clarify the meaning of the terms ‘armed attack’ and indicate that the trend towards a stratified general framework was not fully ‘grounded’ in treaty practice.Footnote ²⁹³ As regards the weight of this ‘other practice’, it is worth reiterating that instances of asymmetrical self-defence were not rare, and that the views of responding States often were set out clearly. The largely negative, or at best indifferent, response to such claims is of course significant; it reduces their probative value significantly. But contrary to common perceptions, claims of asymmetrical self-defence were not systematically rejected. Subsequent practice suggests that they could in principle (though exceptionally) be accepted; there was room for nuance in the Cold War era.

1. Regular, Heterogeneous Practice

The practice of States invoking self-defence against non-State actors is the most obvious indicator of change. Over the course of the past twenty-five years, self-defence claims have become significantly more common and more diverse.

2. A Greater Willingness to Accept Claims of Self-Defence

As was the case during the Cold War era, many of the more recent self-defence claims have been discussed internationally, including in the United Nations. As before, the international community’s response has not been uniform. It has not always clearly focused on the proper construction of the armed attack criterion either: States have often preferred to address proportionality or been guided by overarching perspectives on the conflict.Footnote ³³⁹ Yet these caveats notwithstanding, international reactions confirm a changing attitude: gradual, no doubt, but palpable. The following four considerations illustrate this change.

3. The Appeal of Normative Continuity

Whilst the preceding discussion emphasises elements of change, this change has been accompanied by affirmations of continuity. As in many other disputes about the proper interpretation of the ius ad bellum, there has been a tendency to protect the normative acquis against rupture. This is particularly obvious from assessments removed from the ‘heat’ of a particular dispute.

1. General Considerations

Since 1945, States and UN organs have sought to give meaning to the ‘armed attack’ requirement, in their assessment of particular self-defence claims, and in the course of attempts at clarifying the law in general terms. Their subsequent practice is rich and diverse. For the most part, it has been a debate about nuances – about exploring under what circumstances a State would have to answer for armed attacks carried out by non-State actors. The preceding synthesis of practice highlights that over time, States’ views on this question have changed considerably. And yet, contrary to a popular narrative, this is not a story of radical breaks, let alone of one clear trajectory. If anything, there have been two main trends: first, the move towards a stratified framework in the course of debates about GA Res. 3314, amplified by the Nicaragua judgment and the ILC’s work on State responsibility; and second, the rise of asymmetrical self-defence since the end of the Cold War. This analysis contradicts the popular assumption that practice had been State-centric for decades before the ‘tidal changes’Footnote ³⁹⁷ of early twenty-first-century practice swept aside restrictive approaches: the pendulum has swung back and forth.

What is more, throughout these pendulum swings, the law has remained contested. Even during the heyday of a restrictive analysis, the international community could tolerate the odd instance of asymmetrical self-defence. Conversely, the broad reading of self-defence of the post-Cold War era is primarily ‘driven’ by claim and contestation in concrete disputes, but has not trickled down to the level of multilateral documents; these – just as the ICJ – emphasise continuity. This is, in short, not a simple story of ‘things are different now. Practice, law and/or its interpretation were radically transformed by “9/11” and its aftermath.’Footnote ³⁹⁸ What has changed is the relative support for the different positions.

2. The Significance of the Two Main Trends

The significance of subsequent practice for the interpretation of Article 51 to some extent depends on these preliminary considerations. More importantly, though, it depends on a detailed assessment of the two main trends just identified. This assessment needs to reflect the general rules governing the role of subsequent practice in treaty interpretation, as reflected in the ILC’s recent work.

Looked at from that perspective, the first of the two main trends – the stratification of the regime during the Cold War era – at the time undoubtedly was significant. GA Res. 3314 reflected a ‘subsequent agreement’ of UN members to view self-defence primarily as an inter-State defence. This agreement was recorded in an important document, through which the parties purported to clarify the meaning of aggression, and by extension, self-defence. The key provision of that agreement, Article 3(g), was debated at length; this suggests that it should have significant weight.

For the purposes of treaty interpretation, Article 3(g) offered three main messages: (i) there was some room for asymmetrical self-defence; (ii) such asymmetrical self-defence was to be exceptional – as a minimum, it presupposed a State’s ‘substantial involvement’; and (iii) this by implication seemed to exclude asymmetrical self-defence on the basis of the harbouring (and similar) doctrines. Beyond that, however, the precise contours of the ‘substantial involvement’ test remained unspecific – which, in line with the general regime, would affect its weight.Footnote ³⁹⁹

The Nicaragua judgment and the ILC’s work on State responsibility ‘finetuned’ these messages. Nicaragua suggested that State support would be insufficient – and was taken to suggest that attribution was required, which the ILC construed narrowly. Yet these subsequent ‘specifications’ had to be seen against the backdrop of an inconsistent practice, which could raise doubts about whether the very narrow construction of Article 3(g) was firmly established.Footnote ⁴⁰⁰

The significance of the second pendulum swing – the move towards a broader understanding of self-defence in the post-Cold War era – is more difficult to assess. While the analysis has revealed a palpable change in practice, it needs to be asked whether this change could neutralise the earlier move towards State-centrism. The question is pertinent as the broader construction of self-defence in recent practice is not reflected in general documents of the calibre of GA Res. 3314; in the ILC’s terminology, it is relevant only as ‘other subsequent practice’.Footnote ⁴⁰¹ Some commentators suggest that this ‘lesser’ practice could not modify the earlier position.Footnote ⁴⁰²

At least for the purposes of treaty interpretation, such an approach seems too hierarchical and formal. Subsequent practice is only one factor to be ‘taken into account’ in processes of treaty interpretation; and its weight depends on material criteria (specificity, clarity, repetition). Commentators requiring something akin to an actus contrarius for changes to take effect fail to appreciate the character of interpretation as a ‘single combined operation’.Footnote ⁴⁰³ They also ignore the terms of GA Res. 3314, which – far from purporting to ‘freeze’ the law – was not meant to ‘diminish[] the scope … [of the] Charter … provisions concerning cases in which the use of force is lawful’.Footnote ⁴⁰⁴ All this suggests that the impact of the more recent ‘other subsequent practice’ cannot be dismissed a limine, but needs to be carefully assessed.

3. In Particular: The Impact of the Post-Cold War Practice

A careful assessment needs to inquire whether developments since the end of the Cold War are significant enough to challenge the impact of the State-centric trend of the 1970s and 1980s. This question eschews a clear-cut answer. The preceding analysis suggests a nuanced response that distinguishes between the different ‘messages’ identified above and approaches the problem in engaging with three questions.

1. The Meaning of the Text: Reprise

Given ongoing uncertainties about the proper source of self-defence, section III of the present study has engaged with a range of textual, contextual, teleological and historical arguments. Contrary to a popular understanding, these arguments on balance point towards a broad understanding of the ‘armed attack’ requirement that leaves room for asymmetrical self-defence. The terms of Article 51 are particularly significant in this respect: the provision refers to ‘armed attacks’ without specifying the character of the attacker. This is to be taken at face value, not only because the text is ‘presumed to be the authentic expression of the intentions of the parties’,Footnote ⁴¹⁸ but also because Article 51 explicitly qualifies the character of the victim (which must be a State): ‘what matters’ is indeed ‘the armed attack, not the attacker’.Footnote ⁴¹⁹

Contextual arguments on balance reinforce this understanding. Notably, over time, the Charter’s other recognised exception to the ban on force – the use of force with Security Council authorisation – has come to be read asymmetrically: under the Charter’s collective security system, States can clearly be authorised to use force on the territory of another State even where that other State has not committed any wrong (let alone violated Article 2(4)); the State-centric approaches of the early Charter era have been overcome. Self-defence could be construed in much the same way. Contrary to a prominent view, it need not be read as the ‘flip-side’ of the State-centric ban on force. All this suggests that, when construed in light of the general rule of treaty interpretation, Article 51 is open to, and actually encourages, a reading that permits self-defence against armed attacks by non-State actors.

2. Subsequent Practice: Reprise

This finding crucially affects the role of subsequent practice in the process of treaty interpretation. Embedded in a broader inquiry, subsequent practice is one among a range of elements of interpretation, which ‘in [its] interaction with other means of interpretation, [contributes] to the clarification of the meaning of a treaty’.Footnote ⁴²⁰ The assessment of subsequent practice, in other words, does not take place in a legal vacuum, and nor does it start from scratch. Part of the ‘single, combined operation’, subsequent practice is assessed with a view to determining whether the parties’ conduct in the application of a treaty ‘confirms or modifies the result arrived at by the initial interpretation of the ordinary meaning (or by other means of interpretation)’.Footnote ⁴²¹ Applied to the present case, it is thus to be asked whether the subsequent practice of States and UN organs in the application of Article 51 ‘confirms or modifies [the potentially “open” construction of self-defence] arrived at by the initial interpretation’.Footnote ⁴²²

Looked at from this vantage-point, the long synthesis of subsequent practice conducted in section IV can be distilled into five main points:

1. (i) During the UN’s early practice, States and UN organs seemed prepared to make some use of Article 51’s potential for openness: in debates about indirect aggression, the possibility of self-defence against ‘private’ armed attacks that another State had supported or allowed to be conducted from its territory was discussed.

2. (ii) The General Assembly’s Definition of Aggression constrained options for such flexibility: in it, all States agreed on a more State-centric approach that permitted asymmetrical self-defence only in exceptional settings. The flexible formulation of the exception (‘substantial involvement’) seemed to leave some leeway, though.

3. (iii) In its Nicaragua judgment, the International Court of Justice interpreted the ‘substantial involvement’ test very narrowly and largely deprived it of meaning. The Nicaragua approach effectively construed self-defence as a right to respond against armed attacks by another State. This approach – corroborated by the agreement on a narrow concept of attribution in the ILC’s work on State responsibility – excluded the possibility of asymmetrical self-defence and therefore, at the time, pointed towards a modification of the results of the ‘initial interpretation’ of Article 51. It was, however, not strictly followed in the actual practice of States, which tolerated the odd instance of asymmetrical self-defence.

4. (iv) In the post-Cold War practice, the potential for asymmetrical self-defence is being reclaimed. The palpable change in practice is not so far recorded in framework texts or international jurisprudence of the calibre of the Definition of Aggression or the Nicaragua judgment. However, it is reflected in a rich body of subsequent practice that gives expression to the views of a large number of States. Insofar as these States, as a basic proposition, claim that there must be some room for asymmetrical self-defence outside cases of attribution, their views meet with little resistance. The more specific claim that asymmetrical self-defence should be available against States that had actively supported private armed attacks also has relatively broad support. In both respects, the recent subsequent practice neutralises the earlier, more restrictive approaches – and returns to the result of the ‘initial interpretation’.

5. (v) Insofar as the recent practice supports a right of self-defence on the basis of broader doctrines (harbouring, loss of effective control), it meets with greater resistance. This does not mean that self-defence has been categorically excluded. Quite to the contrary, self-defence claims based on these doctrines have been tolerated or widely endorsed in a relevant number of individual instances. However, there is as yet no clear pattern suggesting under which circumstances they would be tolerated or endorsed. In the absence of clear guidance, all that can be tentatively said is that States relying on self-defence have to be particularly careful to establish the necessity of their action and in particular set out why the armed attack cannot be repelled or averted by the host State.

***

This summary suggests that over seven decades of subsequent practice under the Charter, States and UN organs have given meaning to the ‘armed attack’ requirement. Notwithstanding its trends, twists and turns, the subsequent practice in a number of ways helps concretise the scope of Article 51. It does so by specifying the nexus required to make a State answerable for (or requiring it to tolerate responses directed against) armed attacks by private individuals.

In two respects, the subsequent practice confirms the findings of the initial interpretation of Article 51. It recognises that there is in principle room for self-defence against non-State actors: this has been the position during most of the Charter’s history, overshadowed only during the dominance of the Nicaragua judgment; and it clarifies that States supporting ‘private’ armed attacks can be targets in self-defence.

Beyond instances of State support, matters are more complex. The subsequent practice provides conflicting guidance. While there is much resistance against explicitly recognising self-defence against States harbouring non-State actors, or against States that fail to exercise control over non-State actors operating on their territory, in practice such a right is gradually being accepted in individual instances. Practice certainly no longer excludes self-defence based on a more tenuous State nexus categorically; but it offers as yet neither strong support nor clear guidance.

There are different ways of describing the resulting uncertainty. With Tom Ruys, one can fall back on a trusted formula developed in British government parlance and describe self-defence on the basis of the ‘harbouring’ and ‘loss of effective control’ doctrines as ‘not unambiguously illegal’.Footnote ⁴²³ A more cautious reading would require change to be consolidated before it produces legal effects: during the ‘interregnum’,Footnote ⁴²⁴ the old law would then govern by default.Footnote ⁴²⁵

A more convincing approach ought to proceed from the result of the initial inquiry, which after all points towards an asymmetric reading. In that perspective, one might view in the handling of individual cases since the end of the Cold War a willingness to ‘activate’ an option that was there from the beginning. This shift of perspective results in a slightly different verdict: The ‘harbouring’ and ‘loss of effective control’ doctrines no longer appear merely ‘not unambiguously unlawful’ – but have been accepted in individual cases and thus are no longer excluded in principle. And they have been so accepted, not as ‘novel rights’,Footnote ⁴²⁶ but on the basis of a treaty clause that has, since its entry into force, permitted, and indeed encouraged, asymmetrical readings of self-defence.

***

This final consideration yields a broader lesson and permits us to return to a point made at the outset. As noted in the Introduction, the preceding analysis is a narrow one; its exclusive focus has been on the threshold question, viz. can self-defence be invoked against attacks by non-State actors? Responses to this question are always ‘in principle’ responses: they clarify against which types of attacks the right of self-defence can be invoked, but do not imply that in a particular instance all the conditions for the lawful exercise of self-defence had been met. Whether such exercise has been lawful not only depends on the intensity of the prior attacks (for they have to qualify as ‘armed attacks’, which according to the dominant reading presupposes a certain gravityFootnote ⁴²⁷), it also depends on a host of contextual factors, among them the strength of the reacting State’s factual case and the persuasiveness of its claim that military action is necessary as a measure of last resort and that it be exercised on the territory of the targeted State. These contextual factors have not been the focus of the present inquiry. Yet the preceding analysis of subsequent State practice suggests that they determine whether self-defence can exceptionally be invoked on the basis of doctrines beyond complicity. It indicates that the line between lawful and unlawful exercises of self-defence can not, or no longer, be drawn on the basis of a pre-defined State nexus (‘harbouring’, ‘support’ etc.), but requires a more comprehensive assessment of the situation. The State nexus certainly forms part of this overall assessment. However, it forms a part only.