Revisiting Coercion as an Element of Prohibited Intervention in International Law

Abstract

International law prohibits states from intervening in the internal and external affairs of other states, but only if the method of intervention is coercive. This Article argues that coercion can be understood in two different ways or models. First, as coercion-as-extortion, a demand coupled with a threat of harm or the infliction of harm, done to extract some kind of concession from the victim state—in other words, an act targeting the victim state's will or decision-making calculus. Second, as coercion-as-control, an action materially depriving the victim state of its ability to control its sovereign choices. This may be done even through acts like cyber operations that the victim state is entirely unaware of. The Article argues that many of the difficulties surrounding the notion of coercion arise as a consequence of failing to distinguish between these two different models.

Table of Contents

1. I. Introduction     602

2. II. The Structure of the Non-intervention Rule     606

   1. A. The Prohibition of Intervention in Outline     606

   2. B. The Relationship Between the Prohibition of Intervention and Other Rules     607

   3. C. Interference with Internal and External Affairs (domaine réservé)     609

   4. D. Coercion     612

   5. E. Mental Element(s): Intent of the Intervening State and Knowledge of the Victim State     617

3. III. The Problem of Justification     620

   1. A. The Problem Explained     620

   2. B. Justification Through the Reserved Domain Element     622

   3. C. Enabling Unilateral Law Enforcement?     623

4. IV. The Extortion Model of Coercion     626

   1. A. The Extortion Model in Outline     626

   2. B. Intent and Knowledge     630

   3. C. The Threshold of Coercion     632

5. V. The Control Model of Coercion     640

   1. A. The Control Model in Outline     640

   2. B. Intent and Knowledge     644

   3. C. Threshold of Coercion     646

6. VI. Conclusion     648

I. Introduction

The prohibition of intervention, requiring states to refrain from coercively interfering in the internal or external affairs of other states, is widely recognized as a cardinal rule of customary international law.¹ There is also widespread agreement about the constituent elements of the rule: (1) an interference with a state's internal or external affairs that is (2) coercive in character. As authoritatively interpreted by the International Court of Justice (ICJ) in the Nicaragua case, the principle of non-intervention

forbids all States or groups of States to intervene directly or indirectly in internal or external affairs of other States. A prohibited intervention must accordingly be one bearing on matters in which each State is permitted, by the principle of State sovereignty, to decide freely. One of these is the choice of a political, economic, social and cultural system, and the formulation of foreign policy. Intervention is wrongful when it uses methods of coercion in regard to such choices, which must remain free ones. The element of coercion, which defines, and indeed forms the very essence of, prohibited intervention, is particularly obvious in the case of an intervention which uses force . . . .²

While the existence and authority of the customary rule prohibiting intervention is not today challenged by any state, the content and application of this rule have been disputed since its inception.³ This is particularly the case with the element of coercion. Even though the ICJ regarded coercion as the “very essence” of prohibited intervention, it did not define it. Nor have scholars or states otherwise managed to agree on a definition. Almost forty years after Nicaragua, “the scope of coercion for purposes of prohibited intervention remains particularly undertheorized and underdeveloped,” as Caroline Krass, the general counsel of the U.S. Department of Defense, rightly noted in a recent speech.⁴

My ambition in this Article is precisely to clarify this notion of coercion. On the one hand, its parameters must acknowledge that states interact with each other and exert pressure on one another all the time, for reasons good or bad. On the other hand, “coercion” must inhibit some forms of pressure, drawing a red line beyond which a state's sovereign policy choices would no longer remain free.⁵

A threat or use of military force in violation of Article 2(4) of the UN Charter to affect such choices would indisputably amount to coercion in this sense. When, for instance, Russia (furtively, but reasonably clearly) threatened Ukraine with invasion on account (inter alia) of Ukraine's sovereign choices to align with the West and pursue membership in the European Union and the North Atlantic Treaty Organization (NATO),⁶ and then went ahead with that invasion when its demands were not met, one could think of no clearer example of a state coercing another with respect to its external affairs.

But the content of coercion below the use of force threshold has remained contested. This has especially been the case with measures of political and economic pressure, including unilateral sanctions.⁷ When, for instance, in 2017 Saudi Arabia and its allies imposed a set of adverse measures on Qatar—a so-called “blockade,” which included flight bans, expulsions of nationals, disruption of trade, and even threats to physically sever Qatar from the Arabian peninsula by digging an artificial canal on Saudi territory—Qatar was asked to comply with a series of thirteen demands, which included shutting down the Al Jazeera TV station and severing ties with Turkey and Iran.⁸ Was this interference with Qatar's internal and external affairs “coercive”? Or, when the United States persists in its comprehensive embargo against Cuba, with the goal of changing the nature of Cuba's government and with enormous consequences for the Cuban people, is this El Bloqueo “coercive” in character?⁹ And so forth.

In addressing such questions we are confronted with the undeniable fact that the consolidation of the modern prohibition of intervention in international law, precipitated by a series of resolutions in the UN General Assembly (UNGA), demonstrated substantial disagreement between states on the legitimacy of such measures.¹⁰ Western states in particular have generally opposed UNGA resolutions promoted by developing states that expressly or implicitly asserted that economic pressure was an unlawful form of intervention.¹¹ The constant and consistent resort to such measures by a substantial number of states, accompanied by their opinio juris that such measures are lawful, would seem to negate any effort to render “unilateral coercive measures” (as the UNGA calls them) unlawful on the basis of custom.¹² Even so, it is questionable whether such a conclusion can be squared with any understanding of coercion based on first principles.

The prohibition of intervention has thus suffered from internal tension and uncertainty, as have its elements. In the past decade or so, however, state practice and opinio juris have been responding to a new challenge—the emergence and use of new technologies that enable states to affect each other remotely by cyber means. States have repeatedly affirmed in numerous fora that the “old” law of non-intervention has to apply to this new problem, in essence saying that some types of cyber operations can constitute (coercive) intervention.¹³ In this particular cyber context, states have increasingly been coming out with formal statements—expressions of their opinio juris—interpreting the rule of non-intervention and its constituent elements and applying them to specific examples, such as interferences with elections.¹⁴ They are partly doing so in response to influential academic initiatives to adapt the existing rules of international law to the new challenges posed by cyber, foremost among them the Tallinn Manual.¹⁵ And in doing so, they contribute to the clarification of the rule not only in the cyber context, but also outside it. I should now acknowledge that my own interest in this subject was provoked precisely by this cyber dimension, especially due to my role as one of the editors of the third iteration of the Tallinn Manual, a work-in-progress at the time of writing. But I am firmly of the view that we cannot arrive at some kind of cyber-specific understanding of non-intervention. This rule cannot apply with clarity only in that particular context, if its foundations and application in other contexts remain unclear.

I will argue in this Article that the coercion element of the non-intervention rule can today be clarified and reconceptualized. My argument is partly normative and partly based on recent developments in state practice. I will explain that coercion can be understood through two different, if complementary, models or conceptions. First, the extortion model, under which the coercing state issues a demand to the victim state, and threatens or implements certain harms if that demand is not met. Coercion-as-extortion consists of imposing costs on the victim state, so as to cause it to change its decision-making calculus and policy choices. This is precisely how coercion has traditionally been understood in this context, as “dictatorial” intervention.¹⁶ The key challenge here is whether the threat of adverse consequences below the force threshold, such as economic sanctions, can be covered by this model, and, in particular, whether coercion-as-extortion can include measures that are otherwise lawful.

The second (complementary) approach to coercion is the control model, which conceptualizes coercion as an action by the coercing state that removes the victim state's ability to exercise control over its policy choices. In this model coercion is not about affecting the victim state's decision-making calculus—the victim state's leadership may even be entirely unaware of the actions taken against it—but consists of a material constraint on its ability to pursue the choices that it wanted to pursue. Consider here, for example, a cyber operation against the elections in another country, which may be entirely unrelated to any demands or threats by the coercing state.

In Part II of this Article I will explain the structure of the non-intervention rule. Part III deals with what I will call the problem of justification. Part IV examines coercion-as-extortion, Part V coercion-as-control, while Part VI concludes.

II. The Structure of the Non-intervention Rule

A. The Prohibition of Intervention in Outline

The non-intervention rule that concerns us here is exclusively interstate in nature: it protects and binds only states.¹⁷ Non-state entities have no duty to respect it, although states may be responsible for a breach of the duty of non-intervention if such an entity acts on their behalf, i.e., if the entity's conduct is attributable to a state.¹⁸ Similarly, state measures of coercion against non-state entities such as armed groups, companies, or individuals are outside the scope of this rule, unless such measures are being used to indirectly coerce a state. I will also not be discussing here provisions in the founding instruments of certain international organizations, such as Article 2(7) of the UN Charter, that prohibit the organization in question from intervening in the internal affairs of its members.¹⁹ Rather, I will be addressing solely the non-intervention principle as it applies between states.

The prohibition of intervention is a rule of customary international law.²⁰ It is not expressly stated in the UN Charter apart from Article 2(7), although it is enshrined in regional treaties, such as the Charter of the Organization of American States (OAS).²¹ The codification of the modern non-intervention rule culminated in the 1970 Friendly Relations Declaration,²² a UNGA resolution adopted by consensus, with the express purpose of restating international law rather than political commitments,²³ and supported even by those Western states that opposed other resolutions dealing with non-intervention.²⁴ The ICJ today appears to see the Declaration as wholly reflective of customary international law, and its Nicaragua holding clearly drew upon it.²⁵ Two key paragraphs of the Declaration formulate the non-intervention principle as follows:

No State or group of States has the right to intervene, directly or indirectly, for any reason whatever, in the internal or external affairs of any other State. Consequently, armed intervention and all other forms of interference or attempted threats against the personality of the State or against its political, economic and cultural elements, are in violation of international law.

No State may use or encourage the use of economic political or any other type of measures to coerce another State in order to obtain from it the subordination of the exercise of its sovereign rights and to secure from it advantages of any kind. Also, no State shall organize, assist, foment, finance, incite or tolerate subversive, terrorist or armed activities directed towards the violent overthrow of the regime of another State, or interfere in civil strife in another State.²⁶

B. The Relationship Between the Prohibition of Intervention and Other Rules

Before examining the constituent elements of the non-intervention rule in more detail, it is necessary to consider the relationship between this rule and its cognates, especially sovereignty and the prohibition on the use of force. That relationship depends on several factors: (1) doctrinally, on the point that the different rules have different elements; (2) normatively, on understanding the reasons that make intervention distinctively wrongful; and (3) on the varying levels of stigma that the international community attaches, or should attach, to violations of each rule.

In Nicaragua, the Court noted that there was substantial overlap between the prohibitions on the use of force and intervention, holding that coercion “is particularly obvious in the case of an intervention which uses force, either in the direct form of military action, or in the indirect form of support for subversive or terrorist armed activities within another State [. . . and that these] forms of action are therefore wrongful in the light of both the principle of non-use of force, and that of non-intervention.”²⁷ On the facts, the Court found that the U.S. supply of weapons and training to the contra rebels was a use of force against Nicaragua, but that “the mere supply of funds to the contras, while undoubtedly an act of intervention in the internal affairs of Nicaragua . . . does not in itself amount to a use of force.”²⁸ The Court subsequently concluded that “financial support, training, supply of weapons, intelligence and logistic support [by the United States to the contras], constitutes a clear breach of the principle of non-intervention.”²⁹

Thus, for the Court in Nicaragua, the provision of weapons and training by the United States to the contras was both a use of force and intervention, while the provision of other support was “only” intervention.³⁰ Later in the judgment the Court turned to the principle of respect for territorial sovereignty and found that its effects “inevitably overlap with those of the principles of the prohibition of the use of force and of non-intervention. Thus the assistance to the contras, as well as the direct attacks on Nicaraguan ports, oil installations, etc. . . . not only amount to an unlawful use of force, but also constitute infringements of the territorial sovereignty of Nicaragua, and incursions into its territorial and internal waters.”³¹

The Court took a similar approach in its Congo v. Uganda judgment, in which it concluded that:

Uganda has violated the sovereignty and also the territorial integrity of the DRC. Uganda's actions equally constituted an interference in the internal affairs of the DRC and in the civil war there raging. The unlawful military intervention by Uganda was of such a magnitude and duration that the Court considers it to be a grave violation of the prohibition on the use of force expressed in Article 2, paragraph 4, of the Charter.³²

In sum, the Court's jurisprudence makes it clear that the same act can simultaneously violate more than one rule. It is also clear that not every act of intervention will rise to the level of a use of force.³³

There is an implicit gradation of stigma in the Court's approach—a violation of the prohibition of the use of force is in some sense more serious than a violation of the prohibition of intervention, which is in turn more serious than a violation of territorial sovereignty (as with, for example, the trespass of one state's police or armed forces on the territory of another).³⁴ That violations of the non-intervention rule sit on a midway point of a kind of hierarchy of gravity also seems in line with state positions.³⁵

The hierarchy of stigma notwithstanding, the overlap between force, intervention, and breach of sovereignty is substantial, but not total. While the prohibition of intervention undoubtedly derives from a wider, overarching principle of sovereignty³⁶—that is, a state has a right to be free from outside intervention because it is sovereign³⁷—the purpose of the prohibition is to protect certain sovereign choices from outside coercion even below the force threshold, and even without infringing upon the victim state's territory.

I should acknowledge at this point that there is an ongoing controversy in the cyber context as to whether sovereignty can be violated by another state's cyber operation, which was the unanimous view of the Tallinn Manual 2.0 experts.³⁸ The contrary position is that sovereignty is merely a principle from which specific rules can stem, and that it cannot be violated as such by cyber means. Initially articulated by some U.S. government lawyers,³⁹ this position was subsequently formally adopted by the UK government.⁴⁰ The UK remains isolated on this point—the United States has not officially adopted this view, despite its origins—and, at least so far, all other states commenting on it have argued that sovereignty is a rule, rather than a mere principle. My own personal view aligns with that of the Tallinn Manual experts, but I will not explore this issue further here.⁴¹ The key point for my present purposes is that, were the UK's position to be seen as correct, the absence of a lower-threshold sovereignty rule would entail that the non-intervention rule would correspondingly become more important in regulating state-to-state peacetime cyber operations. And that would, in turn, incentivize loosening the coercion threshold in order to at least to some extent fill the void left by the absence of a sovereignty rule.

Let us now turn to the constituent elements of prohibited intervention.

C. Interference with Internal and External Affairs (domaine réservé)

An act of intervention has to interfere with a state's internal or external affairs. Such affairs are frequently referred to in the literature as the state's domaine réservé (reserved domain), although the terminology varies somewhat, to my mind without material difference.⁴² I will use the term “reserved domain” as shorthand for a state's internal and external affairs.

The reserved domain is the object of prohibited intervention, while coercion is its instrument or method.⁴³ The two elements are cumulative.⁴⁴ In Nicaragua, the ICJ defined the former as encompassing those “matters in which each State is permitted, by the principle of State sovereignty, to decide freely.”⁴⁵ The Court then gave two very broad examples of internal and external affairs respectively: “the choice of a political, economic, social and cultural system, and the formulation of foreign policy.”⁴⁶ The first example clearly drew on the text of the Friendly Relations Declaration, very much shaped by the need for managing coexistence between two ideologically opposed blocks in the Cold War.⁴⁷ In the cyber context, recent state statements have frequently focused on the conduct of elections as an example of internal affairs within the reserved domain, and on matters such as state recognition or membership in international organizations as examples of external affairs.⁴⁸

The Nicaragua Court's reference to permission by virtue of state sovereignty must mean that the reserved domain does not include those matters on which a state cannot decide freely because its choices are limited by international legal obligations.⁴⁹ It is not, to take the most obvious example, for a state to decide freely whether it should subject a portion of its population to genocide. The reserved domain therefore exists within the bounds of international law.⁵⁰ And as international law is ever growing, through the accretion of custom and the adoption and ratification of new treaties, the reserved domain tends to be shrinking.⁵¹ Perhaps the most substantial development in that regard is the growth of international human rights law, which today imposes many limitations on what states can do to their own people.⁵²

That said, the concept of reserved domain should not be understood as referring to those matters in a state's internal or external affairs that are entirely unregulated by international law. Today there are hardly any such matters.⁵³ Rather, internal and external affairs belong to the reserved domain if a state has any measure of discretion within the bounds of international law regarding such issues.⁵⁴ And despite the substantive growth of international law over the past century, there is no doubt that the scope of the reserved domain remains quite broad.⁵⁵

To give a few examples of reserved domain, when Russia demanded that Ukraine abandon any ambitions to join NATO, this clearly interfered with Ukraine's external affairs, since it is the sovereign right of every state to choose its alliances.⁵⁶ (Whether this interference violated the prohibition of intervention depends, as in all of the examples I will give, on whether it was also coercive—my point here is that the first element of the non-intervention rule, an interference with a state's reserved domain, was satisfied). When in 2017 Saudi Arabia, the United Arab Emirates (UAE) and their partners required Qatar to comply with a list of thirteen demands or face the consequences of refusing to do so,⁵⁷ the majority of these demands clearly fell within Qatar's reserved domain. For example, Qatar was required to scale down diplomatic ties and cooperation with Iran, shut down a Turkish military base in Qatar and halt military cooperation with Turkey, and cease contact with the political opposition in Saudi Arabia, the UAE, Egypt, and Bahrain—demands that undoubtedly interfered with Qatar's external affairs. Similarly, Qatar was required to shut down Al Jazeera and various other media outlets it had supported, and to align its military, political, social, and economic policies with the other Gulf and Arab countries—again undoubtedly an interference with Qatar's internal affairs. On all these matters Qatar had the freedom, within the bounds of international law, to choose its own policies. Some of the demands, however, might not have interfered with Qatar's reserved domain, to the extent that they required Qatar to comply with its prior international legal obligations—for example, the demand that Qatar end its alleged interference in the internal affairs of other sovereign states, end its support for alleged terrorist organizations, or pay reparation for loss of life and other financial losses allegedly caused by Qatar's policies.⁵⁸

Similarly, when the United States imposed a comprehensive trade embargo on Cuba, which has now lasted for more than half a century, it has done so with the clear purpose of changing the nature of Cuba's regime.⁵⁹ While developments in international law with respect to the protection of democracy have to be acknowledged, the reserved domain still paradigmatically includes a state's choice of its political and social system, as proclaimed in the Friendly Relations Declaration and affirmed by the ICJ in Nicaragua. There can be no doubt, therefore, that, when taken as a whole, the U.S. embargo against Cuba interferes with Cuba's internal affairs.

When the Trump administration exerted pressure on countries all over the world, including many allies, to compel them not to install Huawei equipment into their 5G telecommunications infrastructure,⁶⁰ there is again no doubt that this interfered with the reserved domain of these states. The choice of using one vendor or another for infrastructure projects is a decision squarely within the internal affairs of each state, as such subject to no international legal obligation. Similarly, when, in response to the December 2018 arrest in Canada of Meng Wanzhou, a Huawei executive and daughter of the company's founder, pursuant to a U.S. request for her extradition on fraud charges, China arrested two Canadian citizens and exerted other forms of pressure on Canada to compel Meng's release,⁶¹ it is clear that this was an interference with Canada's reserved domain, which in principle includes matters of criminal law and extradition on which every state retains a large measure of discretion.

In conclusion, the notion of reserved domain in the non-intervention context is fluid and relative,⁶² but it does not pose too much difficulty. The only truly complex point is to what extent the reserved domain encompasses situations in which measures are used to enforce compliance with existing legal obligations, which I will return to below when discussing whether intervention can be justified. In all of the examples given above, there clearly was an interference with the target state's reserved domain. The much more difficult question for all of these examples is whether the interference was coercive in character, and it is to this question that I will now turn.

D. Coercion

There is no real definition of coercion for non-intervention purposes in any authoritative instrument. As the Dutch government put it somewhat euphemistically, the “precise definition of coercion, and thus of unauthorised intervention, has not yet fully crystallised in international law.”⁶³ In Nicaragua, despite holding that coercion constituted the “very essence” of prohibited intervention,⁶⁴ the Court said nothing about the meaning of this concept in the abstract. In both Nicaragua and Congo v. Uganda, the Court clearly held that coercion was a concept broader than force.⁶⁵ The only example of coercion below the force threshold given by the Court, however, is that of a state providing support such as money or intelligence to a rebel group on the territory of another state.⁶⁶ Yet, in response to Nicaragua's argument that various forms of economic pressure put upon it by the United States—including the suspension of financial aid, a reduction in the sugar quota, and a trade embargo—constituted coercion, the Court observed laconically that “it is unable to regard such action on the economic plane as is here complained of as a breach of the customary-law principle of non-intervention,”⁶⁷ without providing any reasoning for this position. In particular, the Court said nothing about whether and why such measures do not qualify as coercion, although its decision was most likely motivated by the evident disagreement among states on the issue of economic coercion, during the drafting process of the relevant UNGA resolutions and elsewhere.⁶⁸

We are therefore left with the following “easy” cases of coercion in the context of non-intervention: (1) acts involving the use of force contrary to Article 2(4) of the Charter, whether force is used directly by the intervening state or indirectly, through the supply of weapons or training to a rebel armed group; (2) threats of force, equally prohibited by Article 2(4), which are, as I will explain, especially pertinent to the extortion model of coercion; and (3) support for non-state armed groups operating on another state's territory below the force threshold, such as through the provision of funding or the sharing of intelligence. Another easy case relates to what coercion is not: (4) mere persuasion, i.e., an attempt by one state to convince another that a particular course of action is for some reason optimal, which does not involve any threat of harm by the former.⁶⁹ Simple diplomatic engagement with another state, even if it regards matters that clearly fall within that state's reserved domain, such as its choice of a political system or its choice as to what kind of 5 G infrastructure it wants to build, cannot be coercion.⁷⁰

Everything in between these two ends of the spectrum is, in some sense, a “hard” case. A minimalist positivist approach to such hard cases would be simply to deny their hardness, by arguing that coercion should not be understood as a general concept with any distinct content on the basis of some overarching principle, but rather as an enumerated list of actions that states have generally agreed count as coercion, most of which are already prohibited by Article 2(4) of the Charter. In that sense coercion, so it has been argued, would be confined to items (1)–(3) above, and nothing else.⁷¹

An alternative view, which I very much favor, is that a commitment to positivism does not compel this kind of impoverished minimalism, which would reduce the notion of coercion to an enumerated list. If an action is coercive only if states specifically agree that this should be the case, then it becomes impossible to apply this notion to any new development, including cyber operations, absent new state agreement, i.e., some form of lawmaking via treaty or custom, which is highly unlikely to take place with any degree of promptness. To give an example, even though virtually all states that have recently published their formal positions on the application of the non-intervention principle to cyberspace have pointed to some types of election interference operations as a paradigmatic example of a violation of this principle,⁷² the vast majority of states have still not expressed their views on any of these issues. The minimalist position also renders irrelevant the idea that intervention is a combination of two constitutive elements, because there is no need for having these elements if all instances of intervention anyway require specific agreement by states. Moreover, its reduction of coercion to force plus non-forcible support to rebels narrows the scope of prohibited intervention so much that it becomes underinclusive and redundant.⁷³

A minimalist approach to coercion also undermines the generality of legal rules, the notion that they can be applied to factual problems that the legislator did not specifically envisage or provide for. The generality of legal rules, often with a clear “core” of situations to which a rule applies and a much less certain “penumbra,” is perfectly normal both in domestic and international law. A (hypothetical) domestic rule prohibiting the use of vehicles in the park may be clear in most of its applications, but less clear in some—for example, as to whether it prohibits the display of a car mounted on a pedestal in the park, or the use within it of a bicycle or a newly developed (and newly insufferable) electric scooter.⁷⁴ But the fact that the rule is indeterminate in some instances does not affect its generality or binding nature. Similarly, in international law we do not in principle define legal concepts as exhaustive lists of universally agreed upon examples. We do not do this for instance, with the notion of prohibited force in Article 2(4) of the UN Charter, or with the notion of fair and equitable treatment due to foreign investors in bilateral investment treaties, or with the human rights to privacy or expression. These are all generic concepts accepted by states as such, even if they are open-textured, with disagreement on some of their specific applications.⁷⁵

In sum, a positivist conception of coercion need not be minimalist and confined only to those examples of prohibited intervention affirmed as such by the ICJ in Nicaragua. A positivist can argue, as I do, that the prohibition of intervention is a rule of customary international law, but that this rule and its elements such as coercion exist at a higher degree of abstraction or generality. That is, states concur that coercion is a required element of intervention,⁷⁶ see it as a generic concept, and agree that it is in principle broader than force, but might not see its boundaries identically.⁷⁷ There is nothing unusual in this. Some measure of indeterminacy is just fine, as is managing indeterminacy by utilizing arguments about policy, effectiveness, or justice.⁷⁸

How then should coercion be defined if not as an exhaustive list of universally accepted instances? We can briefly dispense with the dictionary definitions of coercion such as “the act or process of persuading someone forcefully to do something that they do not want to do,”⁷⁹ or as acting “to compel to an act or choice; to achieve by force or threat; to restrain or dominate by force.”⁸⁰ Linguistically coercion is clearly a cognate of force and is an act done with the purpose of changing the victim's behavior. But the meaning of words such as coercion and force in ordinary language is relevant for us only marginally, because of their distinct uses as terms of art in the very specific (and interstate) international legal context. That said, the ordinary meaning of the term still includes forms of compulsion other than force, and so we are essentially left in the same place as after reading the ICJ's judgment in Nicaragua.

For their part, the Tallinn Manual experts defined coercion as “not limited to physical force, but rather . . . an affirmative act designed to deprive another State of its freedom of choice, that is, to force that State to act in an involuntary manner or involuntarily refrain from acting in a particular way.”⁸¹ In a similar vein, the Netherlands, one of the few states to have provided a more general definition of coercion,⁸² has done so as “compelling a state to take a course of action (whether an act or an omission) that it would not otherwise voluntarily pursue,”⁸³ essentially a more succinct version of the Tallinn Manual definition.

These definitions point to the undermining of the victim state's will as the essence of coercion (“involuntary,” “voluntarily”), so as to effect a change in its behavior. This somewhat anthropomorphic approach corresponds to the extortion model of coercion, and is similar to how coercion is conceptualized in many domestic law contexts. But it arguably does not fully capture what coercion can mean in the non-intervention setting, and tells us nothing about what types of actions short of force can count as coercion.

The notion of coercion has given rise to much thinking in moral and legal philosophy.⁸⁴ Coercion also figures prominently in various domestic legal contexts, some of which I will explore further below. International law has a few direct analogues. For instance, in the law of treaties, coercion invalidates state consent to be bound by a treaty in two distinct scenarios. The first, per Article 51 of the Vienna Convention on the Law of Treaties (VCLT), is coercion of a state representative who had expressed consent on the state's behalf: “[t]he expression of a State's consent to be bound by a treaty which has been procured by the coercion of its representative through acts or threats directed against him shall be without any legal effect.” This form of coercion is directed against an individual and is clearly broader than the use of violence (“acts or threats” can include non-forcible compulsion).⁸⁵ The second, per Article 52 VCLT, is coercion of the state as such: “[a] treaty is void if its conclusion has been procured by the threat or use of force in violation of the principles of international law embodied in the Charter of the United Nations.” Here coercion is similar in nature to the one in the non-intervention context, i.e., it is the state as an abstract entity that is being coerced, but the scope of coercion is strictly limited to the use or threat of force contrary to Article 2(4) of the Charter.⁸⁶

Similarly, in the law of state responsibility, coercion can function as a rule for attributing derivative responsibility—if one state coerces another to commit an internationally wrongful act, the coercing state is responsible for that act.⁸⁷ The International Law Commission (ILC) defines coercion in this context narrowly, saying that “[n]othing less than conduct which forces the will of the coerced State will suffice, giving it no effective choice but to comply with the wishes of the coercing State.”⁸⁸ But even with this ostensibly narrow approach, the ILC does not confine its definition of coercion to unlawful force, arguing that the compulsion it had in mind can “involve intervention, i.e. coercive interference, in the affairs of another State”⁸⁹ and that “coercion could possibly take other forms, e.g. serious economic pressure, provided that it is such as to deprive the coerced State of any possibility of conforming with the obligation breached.”⁹⁰ Finally, coercion can amount to force majeure and constitute a circumstance precluding wrongfulness—that is, a state that is coerced by another into committing an internationally wrongful act may escape responsibility for committing that act, for which the coercing state would, however, remain responsible.⁹¹

A common thread here is that threats or uses of force are paradigmatic examples of coercion. But coercion is not generally limited to force and can extend to other forms of compulsion. There is no intrinsic point at which non-forcible pressure amounts to coercion; rather, that threshold is variable and contextual.

How precisely coercion should be conceptualized for the purposes of the prohibition of intervention depends primarily on its distinct purpose in the international legal system. As noted above, the central argument of this Article is that, in the context of the non-intervention rule, coercion can be understood in two different but complementary ways. In coercion-as-extortion, the coercing state issues a demand to the victim state, and threatens or implements certain harms if that demand is not met. In the extortion model, coercion consists of imposing costs on the victim state so as to change its decision-making calculus and cause it to change its policy choices.⁹² By contrast, coercion-as-control is an action by the coercing state that removes the victim state's ability to exercise control over its policy choices. In this model coercion is not about affecting the victim state's decision-making calculus but consists of a material denial of its ability to pursue the choices that it wanted to pursue. The two models will be addressed in greater detail in Parts IV and V below.

E. Mental Element(s): Intent of the Intervening State and Knowledge of the Victim State

While there is widespread agreement that an interference with internal or external affairs (reserved domain) and coercion are constitutive elements of the prohibition of intervention, there is more controversy as to whether the non-intervention rule requires subjective mental elements. There are two such possible elements, each of which are generally argued to exist (or not) by virtue of inherently being part of the notion of coercion. First, there is the question whether coercion, and thus intervention, can only exist if the intervening state intends to coerce the victim state, i.e., whether to be coercive the intervening state's relevant action has to be intentional in some sense of that word. Second, whether, in order to be coerced, the victim state must know of the coercive action of the intervening state.

That coercion must be intentional is often regarded as self-evident.⁹³ Arguments in favor of subjective elements tend to turn on the nature of the notion of coercion, on the alleged impossibility of coercing without intending to do so, or of being coerced without being aware of that fact.⁹⁴ Arguments against these elements claim that it is somehow inappropriate to use subjective elements in prescribing the rules of conduct for abstract legal entities such as states, or to assess their responsibility for breaching these rules.⁹⁵ They also point to practical difficulties in proving such mental elements, whether in or outside a courtroom.

As I will shortly explain, subjective elements will depend on the type of coercion we are discussing. But before doing so, it is important to note that there is no formal legal reason why the non-intervention rule could not incorporate mental elements. There is certainly no presumption in international law against such elements, nor is it true that they cannot be attributed to an abstract entity such as the state. It is perfectly common both in international and in municipal law for the responsibility of natural or legal entities to depend on subjective considerations of culpability (fault).⁹⁶ In some contexts, such as criminal law, such considerations take center-stage, and may be required by fair-labeling concerns and the moral condemnation implied in determining that a breach of the law has occurred. But considerations of fault are not limited to the criminal law context.

As far as the international law of state responsibility is concerned, if a primary rule of international law requires the existence of a mental element, this simply means that a natural person, a human being whose conduct is attributable to the state, has had the required level of fault. For example, for a state to be held responsible for committing genocide, one of its organs or agents had to possess the requisite genocidal intent.⁹⁷ The proof of such mental states of individuals—those who act on a state's behalf—is done routinely in judicial settings and outside them, employing multiple modes of evidence, including inference.⁹⁸ Specific rules of international law can, and frequently do, require states to act, through their organs or agents, with some level of fault, such as intent or knowledge. Whether they should do so depends on the specific context.⁹⁹

The real question, therefore, is whether the customary non-intervention rule requires some degree of fault. That question cannot be answered by relying on default assumptions or presumptions. Turning to the primary authorities on the rule, such as the Friendly Relations Declaration or the OAS Charter, we can establish that some of these texts do use language that implicitly refers to some form of intent. For example, the Declaration stipulates that “[n]o State may use or encourage the use of economic, political or any other type of measures to coerce another State in order to obtain from it the subordination of the exercise of its sovereign rights and to secure from it advantages of any kind.”¹⁰⁰ Similarly, the OAS Charter provides that: “[n]o State may use or encourage the use of coercive measures of an economic or political character in order to force the sovereign will of another State and obtain from it advantages of any kind.”¹⁰¹

In Nicaragua, Nicaragua itself argued that the U.S. measures against it were done with two purposes—overthrowing the Nicaraguan government or alternatively coercing it into accepting American demands,¹⁰² and that these two purposes were plainly evident from numerous official statements by high-ranking U.S. officials.¹⁰³ On this point the Court did not expressly say whether some form of intention was indispensable to the notion of coercion in the non-intervention context, but implicitly this appears to have been its position. Nevertheless, the Court's key holding was that it was unnecessary to prove that the United States specifically intended to overthrow the Nicaraguan government in order to find that it intervened against Nicaragua by supporting the contras. In other words, a coercive intent of the United States to compel Nicaragua to make or abstain from certain policy choices should not be equated with the specific intent of the contras to overthrow the Nicaraguan government.¹⁰⁴ The Court followed the same approach in Congo v. Uganda.¹⁰⁵

In the cyber context, the Tallinn Manual experts were somewhat divided on the question of intent, with the majority being of the view that coercion had to be intentional.¹⁰⁶ As for the recent formal positions of states on non-intervention in cyberspace, some have pronounced on this particular point directly. For example, according to the Dutch government, intervention is defined as “interference in the internal or external affairs of another state with a view to employing coercion against that state.”¹⁰⁷ Further, “[t]he goal of the intervention must be to effect change in the behaviour of the target state.”¹⁰⁸ For Finland, hostile interference by cyber means against a state can constitute coercion if “it is done with the purpose of compelling or coercing that State in relation to affairs regarding which it has free choice.”¹⁰⁹ According to New Zealand, coercion requires “an intention to deprive the target state of control over matters falling within the scope of its inherently sovereign functions . . . . While the coercive intention of the state actor is a critical element of the rule, intention may in some circumstances be inferred from the effects of cyber activity.”¹¹⁰ For Canada, coercion includes cyber activities “designed to deprive the affected State of its freedom of choice.”¹¹¹ For Switzerland, coercion means that “a state seeks to cause another to act (or refrain from acting) in a way it would not otherwise.”¹¹² And according to Germany, “the acting State must intend to intervene in the internal affairs of the target State—otherwise the scope of the non-intervention principle would be unduly broad.”¹¹³

Some states have used language that implicitly rather than explicitly evokes or requires intent, for instance when discussing specific examples of prohibited intervention.¹¹⁴ But many states have stayed silent on whether the prohibition of intervention incorporates subjective elements of fault, even when they opined on other aspects of the rule. This is either because they did not consider the issue to be of sufficient importance, or (more likely) because they were unsure what the optimal position on this point should be and wanted to reserve their view for some later date. None of these states have addressed the question whether the victim state needs to know that it is being subjected to coercion, which had also divided the Tallinn Manual experts (the majority being of the view that there was no such requirement).¹¹⁵

What, then, are we to make of these subjective elements? The first point to make is that, because they relate to the coercive action, these elements are part and parcel of the notion of coercion and do not exist independently of it. This means that the existence and nature of these elements will inevitably depend on the model of coercion being employed. Second, discussions of intent often lead to misunderstanding because that word can be used with different meanings, and because such discussions are often hampered by inconsistent terminology (e.g., intent, purpose, with a view, goal, etc.). Third, in that regard, both the proponents and opponents of an intent element do not actually disagree that the coercive action by the intervening state must be intentional under some minimal description.¹¹⁶ There is no conceivable scenario in which a state wholly inadvertently coerces another. Rather, they disagree on whether a particular effect of the action (the description of which varies) must be intended for that action to be labeled as coercive. This will again, as I will explain shortly, depend on the model of coercion being used. Fourth, whether the coerced state needs to have knowledge of being coerced is an issue easier to settle than that of intent. I will develop this point below, when discussing the two models. Finally, the stakes of this debate are in fact lower than they first appear when it comes to coercion-as-extortion. As I will explain, under the extortion model of coercion (Part IV) the coercing state necessarily intends that its demand and threat cause a change in the behavior of the coerced state, while the coerced state is necessarily aware that it is being coerced. By contrast, under the control model (Part V) the coercing state issues no demand and no threat, and here the issue of intent matters much more than with coercion-as-extortion. If intent is required, it will be an intention of the coercing state to deprive the victim state of its ability to control its reserved domain.¹¹⁷ However, the victim state need not, and often is not, aware that is being coerced in this sense of the word.

III. The Problem of Justification

A. The Problem Explained

This brings us to what I will call the problem of justification. Because of the structure of the non-intervention rule, the interpretation of its constituent elements will be influenced by views (or intuitions) as to whether intervention, or coercion, can ever be justified in some sense. Consider the example of state responses to Russia's aggression against Ukraine and its purported annexation of Ukrainian territory. At least some such responses harm Russia, and by inflicting harm attempt to influence Russia to modify its behavior or to become unable to further engage in it. And at least some such responses must be justified, even though Russia's will is unquestionably being bent through coercion-as-extortion.

This, then, is the central dilemma of the non-intervention rule. On the one hand, the core value that it protects is that states, by virtue of being sovereign (and additionally because of their people's right to self-determination), have a right to make their own decisions without external coercion. Put differently, the law must protect the autonomy of the (relatively) weak from domination by the (relatively) powerful. Similarly, the non-intervention rule must enable the co-existence of states with different ideologies and political systems, and reduce the risk that coercive measures might escalate to war.¹¹⁸

On the other hand, it is not only that the reality of states constantly exerting pressure of all kinds on one another militates against a very broad understanding of prohibited intervention, in an obvious apology v. utopia dynamic.¹¹⁹ It is also that pressure—or coercion—can be used for objectively good ends. The sovereign right of states to make their own decisions cannot be understood as the apex value of the international system. Sovereignty cannot override every other consideration, for example the human rights of the state's own population, which it may be grossly violating.¹²⁰

It is difficult for the non-intervention rule to navigate this tension because, as articulated in its authoritative sources, it is framed in categorical, absolute terms—“[n]o State or group of States has the right to intervene, directly or indirectly, for any reason whatever, in the internal or external affairs of any other State.”¹²¹ As written, this rule admits of no exceptions. There is no in-built balancing test of justified coercive intervention, for example, one that would be necessary and proportionate to achieving some legitimate end. Thus, as Lori Fisler Damrosch has put it, at the core of the prohibition of intervention lies “an unresolved tension between the desire to find some sort of limiting principle and the reluctance to adopt one that might be too limiting.”¹²² This is, again, especially the case with regard to coercive measures that are themselves otherwise lawful, such as the suspension of trade or financial aid, where the question is whether such measures should nonetheless become unlawful by falling within the scope of prohibited intervention, an issue I will explore in more detail below.

The categorical nature of the non-intervention rule means that considerations of justification are packed into arguments about its scope. In particular, states (and scholars) are incentivized to argue that particular actions, such as economic sanctions, are not coercive in character. This is not because they would regard such actions as falling outside some coherent and principled general concept of coercion, but because such actions are frequently used for good ends, while the rule apparently admits of no possibility of justified coercion.¹²³ The notion of coercion is thus unavoidably distorted. It becomes moralized, tied up with consideration of justifiability, from which it becomes impossible to untangle.¹²⁴

Instead of arguing about the scope of coercion, it is possible to rely on justificatory mechanisms external to the non-intervention rule, the general circumstances precluding wrongfulness recognized in the customary law of state responsibility. Through the operation of these circumstances an act of coercive intervention would not be regarded as unlawful. But these will be of very limited utility. Some, such as consent, cannot apply even conceivably—a state that validly consents to an act by another state by definition could not have been coerced into providing that consent.¹²⁵ Coercion and valid consent are mutually exclusive.¹²⁶ Other circumstances precluding wrongfulness, such as duress and necessity, are limited by strict criteria, and rightly so. Intervention could be considered as excused or justified under these criteria only very exceptionally.

B. Justification Through the Reserved Domain Element

A proper understanding of the reserved domain element of prohibited intervention provides a more coherent way of thinking about justified intervention than arguments about the scope of coercion or reliance on circumstances precluding wrongfulness. Recall how the reserved domain is a concept bounded by international law, encompassing those matters in the state's internal or external affairs over which it has a degree of discretion under international law;¹²⁷ or, in the words of the ICJ in Nicaragua, those “matters in which each State is permitted, by the principle of State sovereignty, to decide freely.”¹²⁸ This definition of reserved domain as bounded by international law logically entails that coercive acts with regard to a state's choices that fall outside the reserved domain cannot be intervention. In other words, even though the coercion element of prohibited intervention might be satisfied, the reserved domain element would not be. The coercive act in question therefore could not violate the customary prohibition of intervention, although it may well violate some other applicable rule of international law.¹²⁹

For instance, a cyber operation that prevented the armed forces of a state from spreading direct and public incitement to genocide over social media—say by shutting down the relevant accounts on that network, or by disrupting Internet access—would not violate the prohibition of intervention, since disseminating genocidal propaganda does not fall within the internal affairs of the target state over which it may decide freely.¹³⁰ The state does not have a free choice, within the bounds of international law, as to whether to commit or incite genocide against its own population. That choice is simply not part of the “internal affairs” the state is entitled to have control over; it retains no discretion over the matter. The cyber operation in question therefore cannot violate the prohibition of intervention, even if we qualified it as being coercive in character (here as coercion-as-control). But, depending on how the cyber operation was done, it might violate the target state's sovereignty or some other rule.

Similarly, if we were to accept arguendo that economic sanctions can constitute coercion for the purpose of the non-intervention rule, not every use of such sanctions would constitute intervention. This would only be the case if they were used to coerce a state regarding policies or choices that properly belong within its reserved domain. If, however, the sanctions were directed against policies or choices that fall outside the reserved domain, they would not violate the prohibition of intervention, again even if we regarded them as coercive.¹³¹

This would of course only be the case if, on an objective inquiry, the particular policy regarding which a state was being coerced was not one on which international law left it with a measure of discretion.¹³² This plainly may be a contested matter, especially in the absence of a binding judicial determination. Similarly, there may be rules of international law that compel a state to achieve a particular result, but leave it with a substantial degree of discretion on how that result is to be achieved. But if this understanding of the reserved domain element is correct—and I fail to see how the commission of internationally wrongful acts could reasonably be said to be within the scope of a state's internal or external affairs—then this element can have a more important role to play in accommodating justifiability considerations within the non-intervention rule than has been acknowledged so far. And it can do so without distorting the notion of coercion. Coercion would not be prohibited by the non-intervention rule so long as it was strictly confined to compelling the target state's compliance with its international legal obligations.¹³³ Again, the key point here is whether, as an objective matter, the target state retains a measure of discretion regarding particular policies or choices.

C. Enabling Unilateral Law Enforcement?

There are several possible objections to the understanding of the reserved domain element given above, all reflecting a concern that such an understanding would (arguably undesirably) enable unilateral coercive measures of law enforcement. First, even if coercive measures are intended to enforce compliance with an international legal obligation, there will be cases in which states disagree on the content of these obligations and such disagreement may be objectively reasonable. Second, it is unclear how coercive measures that do not constitute intervention because the reserved domain element is not met fit into the traditional dichotomy between retorsion and countermeasures (reprisals) in international law.¹³⁴ Third, under the conception of the reserved domain element proposed above, coercive measures would not constitute intervention even if they were grossly disproportionate to the ostensibly legitimate aim of enforcing legal obligations that is being pursued. Finally, and relatedly, this would in effect allow states using coercive measures to evade the customary requirements for lawful countermeasures in the law of state responsibility.

It is fair to say that the conception of the reserved domain element I have argued for does, to some extent, enable unilateralism, in the sense that the fact that particular measures are (1) unilateral and (2) coercive would not render them prohibited under the non-intervention rule, so long as they were being used to enforce compliance with international law.¹³⁵ But some degree of unilateralism and self-help is simply a fact of life in a non-hierarchical, decentralized system of international law, and each of the concerns outlined above has a satisfactory response.

First, it is undeniable that states may take different positions on points of law, or that there may be several plausible interpretations of an obligation incumbent upon a state, while an independent body making an objective determination of what the law requires might not always be available. This is also a fact of life in the international system, and this fact alone cannot be a reason for categorically preventing the use of coercive measures or confining them solely to cases of unambiguously clear violations of international law (which will even then likely be disputed by the offending state). Indeed, the exact same issue arises in the context of countermeasures, and in principle the same answer can be given—a state taking coercive measures to ensure that another state complies with its legal obligations does so at the risk of making the wrong assessment.¹³⁶ If the coercing state's understanding of the law ultimately proves incorrect, the reserved domain element would be engaged, and coercion would constitute prohibited intervention. The key point here is that the reserved domain element is engaged whenever the coerced state, as a matter of an objective inquiry, retains a measure of discretion on how to comply with a legal obligation.

Second, the retorsion-countermeasures dichotomy actually poses no issue for coercive measures intended to enforce compliance with prior obligations.¹³⁷ As I explained above, such measures simply do not constitute intervention (with a caveat further explored below), but that does not necessarily mean that they are lawful. They can still violate other rules of international law, such as sovereignty, the law of immunities, human rights, or trade law. If they do violate other rules of international law, such coercive measures would need to comply with the criteria for countermeasures in order to be lawful. If they do not violate any other rules of international law, they would constitute retorsion.

Put differently, if coercive measures are confined to enforcing compliance with prior obligations, the prohibition of intervention is not the benchmark against which the legality of these measures is to be assessed. For instance, coercive measures that cause grave harms to (say) the access to food or medicines by the population of the affected state may be wrongful not because they coerce that state on a matter on which it actually has no right to be free from coercion, but because they infringe on the human rights of its people, or unduly restrict international trade, or violate some other rule. Consider, for example, trade restrictions employed against Russia in order to make it desist from its aggression against Ukraine. Even if such measures crossed the threshold of coercion (and some of them likely do), they would still not constitute intervention into Russia's internal or external affairs. This is simply because Russia objectively has no right to use force against Ukraine and no discretion on the matter, despite its protestations to the contrary. If such measures contravened World Trade Organization (WTO) law or human rights law or some other body of international law, the criteria for countermeasures would need to be applied for any wrongfulness of such measures to be precluded. But, legal or illegal, coercive measures would still not constitute intervention into Russia's affairs on which it may decide freely.

Third, the issue of the proportionality of the harm inflicted by the coercing state on the coerced state for the purpose of ensuring the latter's compliance with the law only arises outside the confines of the non-intervention rule. That such measures do not constitute intervention says nothing about the permissibility of third-party or collective countermeasures, i.e., those employed by a non-injured state, which remains the subject of controversy.¹³⁸ I do not here take any position on this issue. I am solely arguing that such measures do not constitute intervention if the reserved domain element is not met. This does not mean that the criteria for lawful countermeasures¹³⁹ are being evaded,¹⁴⁰ but that they only apply if a rule other than the prohibition of intervention is engaged.

The final question that arises is whether the wrongfulness of intervention can ever be precluded through the reliance on countermeasures, if both the reserved domain element as understood above and the coercion element are met. To my mind, this issue points to the centrality of intent as the link between coercion and reserved domain, between the threat and the demand.¹⁴¹ As explained above and further explored below, in coercion-as-extortion the coercing state must intend to compel the victim state to comply with its demand. If that demand is compliance with a prior international legal obligation, and that obligation exists as a matter of an objective inquiry, the prohibition of intervention simply does not apply.

In coercion-as-control, however, the coercing state performs an action depriving the victim state of its ability to control its reserved domain, while intending to deprive it of such control. It is that act alone, without any threat or demand, that constitutes intervention. It is possible for such acts of intervention to be used as countermeasures, if they complied with the relevant conditions in the law of state responsibility. Consider a scenario in which State A intervenes in State B by funding a rebel group on B's territory. B responds symmetrically by funding a rebel group on A's territory (an act which, as we shall see, falls within the coercion-as-control model). B then informs A that it will stop the funding if A ceases its own wrongful act of continuing to support rebels on A's territory. To my mind, it is only in a scenario such as this one that intervention can genuinely constitute a countermeasure, so that its wrongfulness is precluded if the relevant criteria are met.¹⁴² In this scenario, B's first-order intent is to deprive A of control over its reserved domain, by funding a rebel group (which would constitute coercion-as-control). It then has a further, second-order intent to use this wrongful act to induce compliance by A with its legal obligation (which would further constitute coercion-as-extortion).¹⁴³ But, unlike in the reserved domain scenarios above, B's own act of intervention has already been completed because the target state was intentionally deprived of its ability to control its reserved domain, and an illegality really is being used to respond to an illegality. This is again, to my mind, the only scenario in which countermeasures can be relied on to preclude the wrongfulness of intervention.

IV. The Extortion Model of Coercion

A. The Extortion Model in Outline

Let us now turn to the extortion model of coercion, which is in many ways paradigmatic of that notion. As we will see, the parameters of that model are broadly similar to how coercion is understood in ordinary language and domestic contexts, as well as in the philosophical literature.¹⁴⁴

In coercion-as-extortion, the coercing state performs at least two, and sometimes three, distinct actions.¹⁴⁵ First, it makes a demand of the victim state, and this demand relates to matters within the victim state's reserved domain, asking that state to engage or not engage in certain behavior. Second, it threatens the victim state with harm (e.g., force, economic sanctions, or cyber operations—the harm itself need not have any connection to the reserved domain, i.e., may deal with an entirely unrelated area) if the demand is not met.¹⁴⁶ It is precisely the threat of harm and its link with the demand that is the essence of coercion: “do what we say, or else.”¹⁴⁷ On its own a demand can never be coercive as such. Third, if the demand is not met, the coercing state may implement the threatened harm in order to induce the coerced state to comply.

For its part, the victim state can either accept or reject the demand. Its decision on this point can shift during the various stages of this demand-threat-harm process—it might accept the demand immediately after the threat is made, so that the harmful consequences are never implemented; it may reject it initially but accept it subsequently, because the harm inflicted upon it has become so great that it concedes in order for it to stop; or, it might never give in to the demand even if the threatened harmful consequences are implemented fully.

In the extortion model, coercion works by the threat or infliction of harm affecting the coerced state's decision-making calculus, that is whatever cost-benefit analysis its leadership might engage in when making their policy choices.¹⁴⁸ The victim state in effect still has a choice as to whether to comply with the demand or not, but that choice is heavily constrained by outside pressure.¹⁴⁹

The examples of the Cuban embargo by the United States, the Saudi-led blockade of Qatar, and Russia's invasion of Ukraine that we have looked at above can all illustrate the operation of the extortion model of coercion. In all of these cases, the coercing state made demands of the coerced state regarding matters in its reserved domain (e.g., changing the nature of the regime, demilitarization, “denazification,” implementation of specific policies), coupled with threats of harm. In all of these cases, the demand was rejected, and the harmful consequences were implemented fully. In none of these cases, however, was the implementation of these harms successful in producing the desired change in the target state's behavior. In the case of Qatar, the Saudi-led coalition eventually gave up on its demands, which Qatar successfully resisted, so that some semblance of normal relations was restored.¹⁵⁰ In the cases of Cuba and Ukraine, the coerced states continue to resist, while the coercing states continue, at the time of writing, to implement the harms. And with regard to both Cuba and Ukraine, the coercing states persist in their demands, while saying that the implemented harms would cease if the demands were met. For example, Russia has declared that it is willing to stop its “special military operation” if Ukraine agreed to demilitarize, “denazify,” and accept Russia's annexation of parts of its territory—as the Russian foreign minister put it, either Ukraine will give in to these demands or “the Russian army [will] solve the issue . . . . They may stop senseless resistance at any moment.”¹⁵¹ Ukraine so far has no intention of doing so.

Bearing this in mind, we can distinguish between four different variants of the extortion model of coercion:

1. (1) Failed coercion through threat alone. State A issues a demand and a threat to State B. B rejects the demand, but A never implements the threatened consequences.

2. (2) Successful coercion through threat alone. State A issues a demand and a threat to State B. Fearing the harmful impact of the threatened consequences, B accepts A's demand before any harm is inflicted.

3. (3) Successful coercion through the infliction of harm. State A issues a demand and a threat to State B. B rejects the demand, and A therefore implements the threatened consequences. Their impact is so severe that eventually B accepts A's demand. A then stops the harm it had implemented.

4. (4) Failed coercion despite the infliction of harm. State A issues a demand and a threat to State B. B rejects the demand, and A therefore implements the threatened consequences. Regardless of their impact, B persists in its rejection of A's demand. A either continues or even escalates the harm it has been inflicting, or eventually gives up as it considers it to be futile.

The four scenarios of coercion-as-extortion above are valid regardless of how we set the threshold of coercion. That is, they can equally apply to threatened uses of force, economic embargoes or cyber operations. A key question that these scenarios pose is whether they would all constitute coercion even at the highest point of threatened or inflicted harm, i.e., force. This, in turn, depends on whether we think that scenarios in which threats alone are made should be treated differently from those in which the threatened consequences are implemented, and on whether we consider that scenarios in which the coercing state succeeds in getting what it wants should be treated differently from those in which it fails.

I would argue that the distinct wrong captured by the prohibition of intervention, and coercion as its constituent element, is in impermissibly affecting the free choices a state is meant to enjoy by virtue of its sovereignty and the right of its people to self-determination, i.e., in the “the subordination of the exercise of its sovereign rights.”¹⁵² The essence of “dictatorial” intervention is precisely in the Diktat that infringes on the victim state's decisional autonomy.¹⁵³ On that logic, inflicting harms on a state to compel its compliance with an illegitimate demand is clearly worse than simply threatening it with such harms,¹⁵⁴ but the threat alone is bad enough.¹⁵⁵ Nor should, on that logic, the existence of intervention depend on whether it is successful in changing the behavior of the victim state.¹⁵⁶ For example, that Ukraine has against all odds managed to resist Russian aggression and demands does not mean that it has not been subjected to intervention. Or, the situation in which Qatar has used its wealth to build up resilience on various fronts, thus managing to resist Saudi demands against it, should not be treated any differently than one in which a weaker state would have found itself unable to resist such demands.¹⁵⁷ Requiring intervention to be successful in order for it to be wrongful would only serve to punish, without any reason of principle, those states that have somehow managed to resist coercion.¹⁵⁸ As the general counsel for the U.S. Department of Defense has rightly put it:

Requiring coercion to actually produce the desired effect could have paradoxical results. It may disadvantage States that are better able to deflect or endure coercive acts because their resilience would foreclose a determination of prohibited intervention. And it could reward States whose attempted intervention fails. It could also mean the same act, directed towards two different States, could be a prohibited intervention against one State but not the other, depending only on attributes of the target State, for example, whether a target State's cyber defenses were advanced enough to withstand the act. Such an outcome could impede the ability of States with more robust and resilient defenses to call out violations and, if desired and appropriate, to respond lawfully with countermeasures or other tools of international law.¹⁵⁹

I too fail to see how scenarios (1)–(4) above could plausibly be differentiated, precisely because doing so would penalize successful resistance against illegitimate threats.

A particularly instructive example of how states today see coercion through the extortion model is a legislative proposal pending before the institutions of the European Union on an EU anti-coercion legal instrument, crafted as a systemic response to arguably unjustified instances of economic coercion against the EU and its member states by powerful third states, such as the United States and China.¹⁶⁰ The proposal for a draft regulation originated with the European Commission,¹⁶¹ and is as of the time of writing being considered by the European Council¹⁶² and Parliament.¹⁶³ Among the envisaged responses to economic coercion is the taking of countermeasures by the EU or its member states, i.e., acts that would otherwise be in breach of their international obligations. But such measures can only be taken if the coercion against the EU or its members is itself wrongful. This is how the proposed regulation defines such instances:

Coercion is prohibited and therefore a wrongful act under international law when a country deploys measures such as trade or investment restrictions in order to obtain from another country an action or inaction which that country is not internationally obliged to perform and which falls within its sovereignty, when the coercion reaches a certain qualitative or quantitative threshold, depending on both the ends pursued and the means deployed. . . . Acts by third countries are understood under customary international law to include all forms of action or omission, including threats, that are attributable to a State under customary international law.¹⁶⁴

While the draft does not expressly invoke the prohibition of intervention, the definition it uses is precisely that of intervention and its two constituent elements—the wrongful act is a violation of the non-intervention rule.¹⁶⁵ Indeed some of the amendments proposed by committees of the European Parliament would include language specifically referring to non-intervention and the Friendly Relations Declaration.¹⁶⁶ Most importantly, coercion here is manifestly conceptualized as a form of extortion. Its definition also adopts the approach to the reserved domain element that I have argued for above, requiring that the object of coercion be to compel action or inaction that the victim state “is not internationally obliged to perform,” and also requiring the crossing of a somewhat vaguely defined threshold of harm.

B. Intent and Knowledge

Let us now return to the problem of the subjective elements of the non-intervention rule, which as I explained above depend on the model of coercion being employed.¹⁶⁷ As we have just seen, in the extortion model the coercing state performs at least two, and sometimes three, distinct actions: demand, threat, and harm. All of these actions must be intentional under one specific description: causing the victim state to comply with the demand, i.e., effecting some kind of desired change in its behavior in line with that demand. Indeed, it can be argued that it is precisely this coercive intent to inflict harm so as to compel compliance with a demand, namely the subordination of another state's will, that captures the distinctive wrongfulness of intervention.¹⁶⁸ The EU definition of coercion we have just looked at is precisely of this kind, and contains an element of intentionality (“in order to obtain”).

As a point of comparison, domestic criminal laws on extortion type of offenses invariably include some kind of intent element, implicitly¹⁶⁹ or explicitly. For example, the Canadian Criminal Code punishes anyone who “without reasonable justification or excuse and with intent to obtain anything, by threats, accusations, menaces or violence induces or attempts to induce any person . . . to do anything or cause anything to be done.”¹⁷⁰ In the law of England and Wales, a person is guilty of blackmail “if, with a view to gain for himself or another or with intent to cause loss to another, he makes any unwarranted demand with menaces.”¹⁷¹ In the U.S. Model Penal Code (and in the jurisdictions which have relied upon it), a “person is guilty of criminal coercion if, with purpose unlawfully to restrict another's freedom of action to his detriment, he threatens to [commit various proscribed actions].”¹⁷² And in the Serbian Criminal Code both the offenses of extortion and blackmail require the offender to be acting with the “intent to acquire material gain for himself or another.”¹⁷³

Similarly, in the extortion model the coerced state will invariably know that it is being coerced, because coercion in this model works through the exertion of pressure on the victim state's decision-making calculus, so that the threatened or actual infliction of harms on the victim state would cause it to change its decisions.¹⁷⁴ Without knowing of the demand or of the threatened consequences, the victim state cannot decide to modify its behavior as a consequence of the coercion.

Proving intent and knowledge in coercion-as-extortion—even if such a requirement formally existed—would generally not be difficult. This is because the demand and the threat need to be communicated to be effective. When, for example, Saudi Arabia and its allies made their list of thirteen demands to Qatar and threatened adverse consequences in case of non-compliance, it is manifest that by doing so they intended to cause a change in Qatar's behavior and it is equally manifest that Qatar knew that. And when their demands were rejected, and Saudi Arabia and its allies implemented the threatened harms, there can again be no doubt that these harms were intended to induce a change in Qatar's behavior, i.e., to compel it to comply with the demands, and that Qatar knew this. Proving coercive intent and knowledge in this extortion context requires little effort. What else could all the relevant actions (demand, threat, harm) be if not intentional, and how could Qatar be coerced unknowingly?

It is true that in some cases the coercing state's threats might be veiled, in the sense that it wants to maintain some kind of plausible deniability and be able to argue that it never made a threat linked to its demands. For example, up until the very moment of Russia's full-scale invasion of Ukraine in February 2022, Russian officials denied that they were threatening Ukraine with war.¹⁷⁵ But the language they otherwise used could easily be understood by everyone as a threat of force, especially in the context of Russia amassing its military on Ukraine's borders. Once the invasion began, of course, any doubt about the nature of Russia's threats were dispelled. In short, while there might be some genuinely difficult cases in which it is unclear whether a state is making threats against another in order to induce it comply with its demands, the very nature of coercion-as-extortion means that proving the existence of such threats will generally be straightforward.¹⁷⁶

This brings us to the first liminal case in which the intent question does matter—let us call it the scenario of successful coercion by bluff. Here the coercing state makes its demand and makes its threat, but it has not yet followed through with the threatened harm. The victim state may nonetheless feel compelled by the threat to give in to the demand and does so. In such a situation it may be perfectly possible that the coercing state was in fact bluffing, i.e., that it never really intended to carry its threat out. Consider a counterfactual scenario—which many at the time thought reflected reality—of Russia amassing its armed forces on the Ukrainian border in late 2021 in order to compel Ukraine to comply with its demands on demilitarization and the like. Yet, in this counterfactual scenario, President Putin was bluffing and never intended to launch an invasion, but nonetheless succeeded in getting his demands fulfilled. At the time, Ukraine, for its part, might not have known, i.e., have been virtually certain,¹⁷⁷ whether Putin was bluffing, depending on the intelligence it had available.

Would we in this scenario regard Ukraine as being coerced into giving in to Russia's demands? The answer surely must be in the affirmative. Properly understood, the threat was intentional, even if it was a bluff. The intention that matters is to cause a change in the victim state's behavior, i.e., to compel it to do or not do something it would otherwise not have willingly done. And in this scenario, it is clear that Russia's demand and threat were intentional in that sense. Whether Russia further intended to carry out its threat is immaterial, because it is the threat alone that could plausibly cause a change in Ukraine's behavior by constraining its will in making its sovereign choices. We could not sensibly require a victim state in this position to positively know that the coercing state is not bluffing, so long as its belief that the threat was genuine was objectively reasonable. In other words, by making a threat, the coercing state can intend to cause a change in the behavior of the coerced state even without intending to carry out its threat in case of non-compliance. The extortion model of coercion requires only the former kind of intent, not the latter. And as I explained above, the extortion model should include scenarios in which the coercing state succeeds in obtaining what it wishes using threats alone.¹⁷⁸

C. The Threshold of Coercion

The key point of uncertainty with regard to the extortion model of coercion is the threshold of the harm that is threatened and possibly implemented by the coercing state. We have already seen above (Section II.D) how there are some easy cases of coercion and not-coercion. A situation in which the intervening state threatens or uses force or supports rebels clearly is coercion. A situation in which there is no threat of harm, but merely an effort of persuasion, cannot be coercion. All other types of threatened or implemented harms in between these two ends of the spectrum are in some sense a hard case, primarily because states have so far been unable to agree on any clear threshold of harm.

Would, for example, threatening or implementing trade restrictions or other economic sanctions cross the threshold of harm so as to constitute coercion? What about threats to suspend military or intelligence cooperation? What about threats to suspend visa-free travel for the other state's citizens or to withhold consent to the state's admission into an international organization? Consider, for example, how for many years Greece obstructed (North) Macedonia's efforts to join the European Union and NATO on account of its dispute about the latter's name,¹⁷⁹ or how Turkey has (as of the time of writing) conditioned Sweden joining NATO on compliance with various Turkish demands, including the extradition of a hundred or so supposed “terrorists.”¹⁸⁰ Is this coercion, or are these “hardball” tactics that should be considered legitimate in international negotiations?

Or consider cases of so-called “hostage diplomacy,” in which one state detains, prosecutes or otherwise ill-treats the citizens of another in order to induce the latter to comply with a demand. When, for example, China detained two Canadians on apparently trumped-up charges in order to induce Canada to release Huawei's Meng Wanzhou,¹⁸¹ was this not only a violation of the human rights of the detained individuals, but also China coercing Canada, and thereby intervening in its internal or external affairs? When Iran detained two British-Iranians in order to compel the UK to pay its (undisputed) debt of 400 million pounds to Iran,¹⁸² was this again not only a violation of these individuals’ rights, but also Iran exercising coercion against the UK?¹⁸³

The question before us, therefore, is whether there can be any threshold of threatened or inflicted harm that can properly be qualified as coercion that is not arbitrary and is not manifestly overinclusive or underinclusive. It should come as no surprise that there are many reasonably analogous questions in domestic law in which the same issue arises, and which are to some extent instructive for discussing coercion in the non-intervention rule. In particular, domestic legal systems not only punish extortion or racketeering-type offenses that involve threats or use of violence in order to obtain money or some other property, but also criminalize other types of coercion that do not involve violence.

For example, under the German Criminal Code the crime of coercion (Nötigung) covers anyone who “unlawfully, by force or threat of serious harm, compels a person to do, acquiesce to or refrain from an act.”¹⁸⁴ Such coercive acts are unlawful “if the use of force or the threat of harm is deemed reprehensible in respect of the desired objective.”¹⁸⁵ The notion of threats of serious harm is considerably wider than force and can include, for example, the unjustified filing of criminal complaints against an individual.¹⁸⁶ Similarly, in Swiss law the crime of coercion covers any person “who, by the use of force or the threat of serious detriment or other restriction of another's freedom to act compels another to carry out an act, to fail to carry out an act or to tolerate an act.”¹⁸⁷ This definition is again plainly broader than force. Similarly, the U.S. Model Penal Code's definition of criminal coercion covers threats to commit any criminal offense (i.e., not just a violent one), as well as threats to accuse anyone of a criminal offense, expose secrets, or take or fail to take official actions.¹⁸⁸ Again, the basic point that emerges here is that when domestic systems define extortion or coercion-type offenses they do not generally limit the threatened harm to violence alone, but do so more broadly, while engaging in some kind of line-drawing exercise.

A particularly instructive domestic example is the crime of blackmail. While the formulation of this offense of course varies comparatively,¹⁸⁹ its distinctive feature is generally that its component elements are not themselves criminal, or not even necessarily immoral. Consider, for instance, a scenario in which someone discovers that an acquaintance is engaging in adultery. Asking that acquaintance for money or some other favor would not in and of itself be either unlawful or immoral. Truthfully informing that person's spouse of the adultery would certainly be lawful and may even be virtuous. But combining these two acts—threatening the adulterer with exposure of his wrongdoing unless he paid a sum of money or complied with some other demand—would constitute a criminal offense in many legal systems and would generally be regarded as immoral. The same would apply, for instance, in a situation where one person threatened another to report her for committing a criminal offense (which did in fact happen), unless the latter paid money or complied with some other demand.¹⁹⁰ This supposed “paradox” of blackmail—that a combination of two legal acts can become unlawful—has been much discussed in the legal and philosophical literature.¹⁹¹ The key takeaway for us here is that, as a legal matter, threatening to commit lawful acts while coupling that threat with a demand can in some circumstances be regarded as unlawful. In that sense, coercion for the purpose of the non-intervention rule can at least conceivably include threats of otherwise lawful actions, e.g., the suspension of some economic relations.

Domestic legal systems deal with many other relationships or transactions that involve a degree of coercion or some form of exploitation. Consider, for example, legal protections of employees from employers; protections of tenants from evictions or exorbitant rent raises by landlords; the rules of both public and private law that deal with the vitiation of consent in situations of particular vulnerability or economic disadvantage, for example in the context of usury, price gouging, form or adhesive contracts, and so forth.¹⁹² In all such cases one party to a relationship or transaction exercises a form of pressure or influence against the other party that is not forcible in nature, but nonetheless genuinely constrains the other party's freedom to make their own decisions or results in substantial unfairness. Some of these cases involve constraint through threats—think, for example, of the situation of an indigent person whose power supplier threatens to cut off their electricity in winter due to unpaid bills, compelling them to choose between heating their home or feeding their family. Others cases may involve constraint through offers rather than threats—for example, a monopolistic large retailer who can simply dictate prices and other contractual terms to companies in their supply chain.¹⁹³

All such cases tend to involve difficult line-drawing exercises between forms of pressure that are considered impermissible, and forms of pressure in “hard” economic transactions that should not be regarded as unlawful, even though they might be unjust or immoral.¹⁹⁴ This is the case even with coercion or extortion-type criminal offenses. Consider, for example, the situation of a union leader who threatens their employer that their workers will go on strike unless their demands for better pay or conditions are not met. That threat, and any industrial action the workers might take, is coercive, but we would widely regard this form of coercion as justified (at least in principle). By contrast, if the union leader threatened the employer with industrial action while demanding money for his or her own personal gain, we would regard such demands and threats as impermissibly coercive, potentially even criminal.¹⁹⁵ Note, for instance, how Canadian law defines extortion as threats made “without reasonable justification or excuse,”¹⁹⁶ how the German definition of prohibited coercion explicitly incorporates a balancing test by providing that threatening harms is only unlawful if it “is deemed reprehensible in respect of the desired objective,”¹⁹⁷ how English law prohibits only (subjectively) unwarranted demands in blackmail,¹⁹⁸ or how the U.S. Model Penal Code expressly provided for an affirmative justification defense for any threatened harms that do not rise to the level of a criminal offense.¹⁹⁹

What can this detour into domestic law teach us about the extortion model of coercion in the international law prohibition of intervention? First, that while focusing on the threat or use of violence is perfectly sensible in understanding coercion, this notion simply cannot be reduced to force alone. If the interest we are protecting is the autonomy of individuals, or states, to make their own decisions, that autonomy can be infringed by multiple methods that can be as effective, and in some circumstances as condemnable, as force. Second, that all domestic systems engage in attempts at line-drawing between permissible and impermissible forms of pressure that suffer from some degree of arbitrariness and indeterminacy, just like in international law—but these are attempts we can learn from. Third, that in many domestic systems even threats of otherwise lawful harms can become unlawful coercion. Finally, that in many contexts the boundaries of impermissible coercion that are drawn by domestic law are shaped by what I termed the problem of justification, i.e., that some instances of coercion can be regarded as justified.²⁰⁰

How, then, can we conceptualize the coercion-as-extortion threshold for the purpose of the non-intervention rule in international law? There are in my view three viable approaches in that regard, which can to some extent be complementary: (1) coercion as threat or use of force, or any other harm below the force threshold that states can specifically agree on; (2) coercion as the threat or infliction of harm that would breach the rights of the victim state or its nationals; (3) coercion as the threat of infliction of harm that is by its character, scale, and effects particularly serious or severe, even though it might not violate the rights of the victim state. Let us examine each in turn.

Force and minimalism. As we have already seen there is no dispute that the threat or use of unlawful force will meet the coercion threshold. The same goes for supporting rebel groups on another state's territory, or threatening to do so. A minimalist approach would include within the notion of coercion those actions on which states can specifically agree, but would exclude any actions on which there is substantial controversy, such as economic sanctions.²⁰¹ For example, because even the most systematic uses of economic sanctions, as with the American embargo against Cuba, have not generally been accepted as violating the prohibition of intervention, such sanctions should simply not be regarded as coercive in character.²⁰²

I have already explained how that approach is underinclusive. As the general counsel for the U.S. Department of Defense has rightly put it:

Although some may assume that a coercive action must involve force—because, of course, force is inherently coercive—such an interpretation is unduly limited. The prohibition against intervention makes sense as a standalone rule under international law only if it is distinct from the prohibition on the use of force, which means coercion for purposes of prohibited intervention in the cyber domain must include at least some acts below the threshold of a use of force.²⁰³

Violation of rights. The second approach would include within the notion of coercion not just the threat or use of force, but also the threat or commission of any internationally wrongful act by the coercing state against the victim state or one of its nationals.²⁰⁴ Thus, for example, if State A threatened State B with cyber operations that would violate B's sovereignty, or went ahead with such operations (e.g., wiper malware attacks against banks and hospitals) if its demand was not met, the cyber operations in question would not only violate B's sovereignty, but would also constitute coercion and thus a breach of the prohibition of intervention. Or, if A engaged in “hostage diplomacy” against B, such an action would not only breach the rights of these individuals, but also B's right to be free of coercive intervention. Or, to turn to economic coercion, such threatened or implemented measures would constitute coercion for non-intervention purposes if they breached some prior obligation owed by A to B,²⁰⁵ for instance under WTO law, international investment law or bilateral treaties of friendship and commerce,²⁰⁶ or indeed if such measures violated the human rights of B's population, say because of their impact on the enjoyment of the right to food or the right to health.²⁰⁷

The rationale behind this approach is that the line to be drawn between permissible and impermissible forms of pressure by one state on another should align with the rules of international law that already apply between the relevant states.²⁰⁸ This approach is analogous to those philosophers of coercion who regards the essence of coercion as the violation of an individual's rights.²⁰⁹

This conception of the coercion threshold would “upgrade” an act that is already unlawful by labeling it as intervention, due to the coercive purpose with which it is used. In addition to this elevated stigma, the practical added value of this approach is that it would render unlawful the mere threats of the relevant internationally wrongful acts when such threats are coupled with a demand and are made with a coercive purpose, even when such threats would not necessarily be unlawful under the specific substantive rules in question. For example, Article 2(4) of the UN Charter expressly prohibits both the use of force and the threats of force, but there is arguably no rule prohibiting states from threatening to breach another state's sovereignty or a rule of WTO law. If, however, such a threat is coupled with a demand in order to coerce the victim state regarding its internal or external affairs, the threat as such would become unlawful qua intervention. Put differently, if State A threatened State B with a cyber operation against its critical infrastructure that would breach its sovereignty, the threat alone would not as such constitute a breach of B's sovereignty unless it was acted upon. But if the threat was made to coerce B, A would be responsible for breaching the prohibition of intervention even if the cyber operation was not conducted and no breach of sovereignty occurred.

The upside of this approach is that it provides a reasonably clear baseline for the concept of coercion, one grounded not just in the prohibition on the use of force, but also in all other relevant rules of international law. Such coherence between the non-intervention rule and the related network of international rules would arguably enhance its legitimacy and compliance-pull.²¹⁰ The downside of this approach is that it is potentially underinclusive in that, with the exception of the threats point above, the non-intervention rule would only prohibit those acts that are already prohibited under some other rule of international law.

Severity of harm. A third approach would ground coercion not in the violation of the victim state's legal rights, but in the magnitude or severity of the harms the coercing state threatens it with or inflicts upon it. In other words, threatening or inflicting severe harms would be coercive if done for the purpose of compelling the victim state to comply with a demand, even if the harm in question would not be unlawful as such.²¹¹

There is no conceptual problem with the prohibition of intervention sweeping more widely than other rules of international law, i.e., prohibiting behavior that is not already prohibited by some other rule.²¹² As the ICJ has put it, “[t]here can be no doubt that, as a general rule, a particular act may be perfectly lawful under one body of legal rules and unlawful under another.”²¹³ We should recall in that regard that the definitions of extortion-type offenses in numerous domestic systems do precisely that. Blackmail is the paradigmatic example here—as explained above, it generally consists of threatening an individual with an otherwise lawful action of disclosing some information embarrassing to them, combined with an otherwise lawful request for money or some other benefit.²¹⁴ It is the coercive threat-demand combination, coupled with an improper purpose linking the two and its infringement upon individual autonomy, that justifies criminal punishment in such situations. Similarly, we have seen how various domestic systems define extortion offenses by reference to the severity or gravity of the harm threatened or inflicted, without necessarily requiring that this harm would be unlawful if taken in isolation.²¹⁵

Grounding the coercion threshold in the severity or magnitude of the threatened or inflicted harm makes normative sense because coercion-as-extortion works precisely on that basis. The threat or infliction of harm is the reason for the coerced state to act and make or change its decisions. The greater the harm, the greater the constraint, and the easier it is to demonstrate that the threat or infliction of the harm could or did cause a change in the behavior of the victim state. And we already employ similar tests, which are in principle workable in practice, in international law. Consider, for example, how the use of force and armed attack thresholds for the purpose of Articles 2(4) and 51 of the UN Charter are habitually defined by reference to their “scale and effects,”²¹⁶ or how the crime of aggression is defined in the Rome Statute by reference to the “character, gravity and scale” of the underlying act of aggression.²¹⁷ Similar approaches have been advocated for violations of sovereignty in the cyber context.²¹⁸

A severity assessment would necessarily be contextual. It would need to consider various factors, including the capabilities of both the coercing and victim states and any relationships of dependency or vulnerability.²¹⁹ A threat that is severe for San Marino might not be severe for the United States. That said, the fact that the victim state may be able to resist the harm and not give in to the intervening state's demands cannot ipso facto mean that the harm was not sufficiently severe. In other words, we should not confuse the severity of the harm with a requirement that the threat or imposition of the harm actually cause the victim state to change its behavior.²²⁰ If we again take the example of the U.S. embargo against Cuba, there should be no doubt that it would meet any severity threshold (to the extent that economic measures can generally cross that threshold) even though Cuba has managed to resist it at great cost.

The main feature of this approach is that it does not require the underlying harm to be unlawful, so long as it is sufficiently severe. It could thus cover measures of economic coercion that are not contrary to say trade law or investment law, such as the suspension of financial aid in a situation of dependency, so long as their impacts were sufficiently severe. Crucially, in their official positions some states have already clearly relied on criteria such as scale or severity to define the coercion threshold.²²¹ And this makes sense—it is again the severity of the threatened or inflicted harm that, in the words of the German government, ensures that the victim state's “will is manifestly bent by the foreign State's conduct.”²²² Perhaps the most detailed articulation of such an approach so far is the proposed EU anti-coercion instrument, which regards as coercive those harms that cross “a certain qualitative or quantitative threshold [. . . ensuring] that only economic coercion with a sufficiently serious impact or, where the economic coercion consists in a threat, only a threat that is credible” is covered by the instrument.²²³

But this approach may also be criticized for relying on vague notions of severity and for being overinclusive, unrealistically prohibiting a lot of “carrot-and-stick” diplomacy that states habitually engage in. One response to that criticism is that the severity threshold can be set relatively high to address overinclusiveness concerns, especially bearing in mind that the non-intervention rule allows for no exceptions. Similarly, that the vagueness of a severity criterion is indisputable does not mean that it is unmanageable—after all, international law already does this in many other contexts, and we have also seen how similar formulations are used domestically even for criminal offenses. That said, there will undoubtedly be some hard cases of applying this approach. Was, for example, Greece's use of unanimity requirements for joining organizations such the EU or NATO to compel Macedonia to change its name (clearly an issue within any state's reserved domain), severe enough in its impacts on Macedonia to count as coercion?²²⁴

In sum, there are multiple possible approaches to calibrating the coercion-as-extortion threshold. At the barest minimum it includes the threat or use of force, but it can be extended without great difficulty to cover the threat or infliction of harms that are already internationally wrongful on some other basis. The real dilemma is whether coercion-as-extortion should also cover threats or infliction of lawful harms that are nonetheless severe in their potential or real impacts on the coerced state. In my view such severe harms should be covered by coercion-as-extortion, and thus constitute prohibited intervention, if they are designed to compel the coerced state to comply with a demand that infringes on its authority to determine its own internal or external affairs. As I already argued, however, if the reserved domain element is not met because coercion is used to compel a state to comply with its obligations under international law, there would be no prohibited intervention. This to my mind sufficiently addresses concerns that the non-intervention rule would become overinclusive.²²⁵

V. The Control Model of Coercion

A. The Control Model in Outline

This brings us to the control model of coercion. Unlike coercion-as-extortion, coercion-as-control requires no demand or threat. Although the coercing state may still make threats and demands, coercion-as-control does not work by influencing the decision-making calculus of the coerced state. Rather, the coercing state directly interferes with the coerced state's internal or external affairs by taking an action that materially deprives the coerced state of its ability to control its policy choices within its reserved domain. A good starting point for thinking about coercion-as-control is the definition of intervention offered by the authors of Oppenheim: “the interference must be forcible or dictatorial, or otherwise coercive, in effect depriving the state intervened against of control over the matter in question. Interference pure and simple is not intervention.”²²⁶ Incidentally it is precisely this definition of coercion that was recently adopted almost verbatim by the United Kingdom's attorney general.²²⁷

We can again understand the essence of coercion-as-control by reference to domestic examples. Consider, for example, situations in which a state agent arrests or detains an individual, or searches their person or property, or seizes funds from their bank accounts, or blocks access to a website. In all of these situations the individual is, in some relevant sense, being subjected to coercion. The state agent might or might not make threats and demands, but the essence of coercion is not here in compelling the individual to comply with any demands under threat, but in performing an action against their will that is to the detriment of that individual and which deprives the individual of their ability to control their person, liberty, property, or some other interest. These are cases of coercion as much as those of coercion-as-extortion.²²⁸

Or, to turn back to international law, consider the one example of cyber activities that all states (so far) seem to agree are coercive in character and constitute intervention—election interference operations.²²⁹ There are many types of such operations, ranging from hacking a state's election commission, vote counting infrastructure or voting machines, to disseminating disinformation to target a particular candidate or suppress the vote of a certain category of people. I think it is fair to say that not all states would agree that every such operation constitutes coercion (and thus intervention), but they do seem to agree that some such operations do (and no state has expressed any dissent). At the most extreme end of the spectrum of such activities, a cyber operation that changed the voting tally in an election or referendum and changed the outcome of that process must constitute intervention.²³⁰

The coerciveness of such operations cannot be explained by the coercion-as-extortion model. Election interference operations are not done pursuant to a demand coupled with a threat, in order to change the decision-making calculus of the coerced state. Rather, the coercing state through such an operation simply deprives the coerced state of control over its internal or external affairs, namely its political system.²³¹ In other words, accepting that such operations constitute intervention logically entails accepting the model of coercion as deprivation of the victim state's ability to control its own affairs.²³² And while thinking of coercion in terms of extortion comes naturally to us, this is not the only way of conceptualizing coercion—the two models are complementary, not mutually exclusive.

In short, coercion-as-control is essentially “a function of the coercer using power to determine what the target will or will not do.”²³³ Coercion-as-control has not clearly been distinguished from coercion-as-extortion in state practice, jurisprudence, or scholarship. Nonetheless, I would argue that it has sufficient support.²³⁴ First, recall how the Friendly Relations Declaration sets out a universally accepted example of coercion, supporting rebel groups on another state's territory: “no State shall organize, assist, foment, finance, incite or tolerate subversive, terrorist or armed activities directed towards the violent overthrow of the régime of another State, or interfere in civil strife in another State.”²³⁵ In Nicaragua, the ICJ similarly treated the support of armed groups on another state's territory as a paradigmatic example of intervention.²³⁶ Yet this type of coercion was not linked by the UNGA or the ICJ to any threats or demands by the coercing state. The mere fact of such support constitutes intervention, since it deprives the victim state of the ability to control its internal affairs and exercise its sovereign authority over its entire territory. This, in other words, is an example of coercion-as-control, not coercion-as-extortion.

Similarly, consider another widely accepted example of intervention—the premature recognition of a separatist entity as an independent state.²³⁷ According to no less an authority than Hersch Lauterpacht, “[i]t is generally agreed that premature recognition is more than an unfriendly act; it is an act of intervention and an international delinquency.”²³⁸ If State A recognizes a separatist entity in State B as an independent state when that entity does not meet the international legal criteria for statehood (which of course can be contested), the mere fact of such recognition is regarded as intervention in B's internal affairs, even if A made no demand of or threat to B. This means that the act of recognition as such is in some sense coercive, in that A deprived B of its ability to fully control its reserved domain by denying B's claim of authority over its entire territory.

A similar example is the unlawful exercise of enforcement jurisdiction on another state's territory, e.g., through searches or arrests without that state's consent, which is generally regarded as violating the territorial state's sovereignty²³⁹ and (at least in some cases) as violating the prohibition of intervention.²⁴⁰ This can only be if by taking such acts the intervening state directly deprived the victim state of its ability to control its internal affairs.

Support for a coercion-as-control approach can also be found in some of the recent state statements on the application of the prohibition of intervention in cyberspace. For example, the United Kingdom's Attorney General not only adopted the Oppenheim definition of coercion, with its reference to deprivation of control, but also added that “we should be ready to consider whether disruptive cyber behaviours are coercive even where it might not be possible to point to a specific course of conduct which a State has been forced into or prevented from taking.”²⁴¹ And this does make sense—coercion-as-control is not about enforcing compliance with a specific demand as in coercion-as-extortion. Further, for Australia:

Coercive means are those that effectively deprive the State of the ability to control, decide upon or govern matters of an inherently sovereign nature. Accordingly, the use by a hostile State of cyber activities to manipulate the electoral system to alter the results of an election in another State, intervention in the fundamental operation of Parliament, or in the stability of States’ financial systems would constitute a violation of the principle of non-intervention.²⁴²

Similarly, according to New Zealand, a cyber activity will violate the rule of non-intervention if it is “coercive (i.e., there is an intention to deprive the target state of control over matters falling within the scope of its inherently sovereign functions).”²⁴³ For Canada, coercion comprises state “activities [that] would cause coercive effects that deprive, compel, or impose an outcome on the affected State on matters in which it has free choice.”²⁴⁴ And these states and others have in their official pronouncements given examples of coercive activities that do not follow the demand-threat-harm structure of coercion-as-extortion, but can only be explained through a coercion-as-control model.²⁴⁵

There is, in short, sufficient support for the proposition that the prohibition of intervention can be violated by certain types of actions that are not coupled with demands and threats. Such actions coerce the victim state directly, by materially depriving it of its ability to control its reserved domain. They do not coerce by exerting pressure on that state so as to change the decisions of its leaders. Coercion-as-control does not lie on the same spectrum as influence, and should carefully be distinguished from coercion-as-extortion. While they share the ultimate intention of the coercing state to produce a change in the behavior of the victim state, they operate differently.²⁴⁶ For instance, in coercion-as-extortion the threatened or inflicted harm need not have any connection to the victim state's reserved domain as such. Rather, it is a means toward an end, which is compliance with a demand that does affect the victim state's reserved domain. It is not, for example, the economic measures that the United States has taken against Cuba that as such interfere with Cuba's reserved domain. Rather, such measures are used to compel compliance with U.S. demands about regime and societal change, which do interfere with Cuba's reserved domain. By contrast, in coercion-as-control it is precisely the action being taken by the coercing state, such as a cyber operation, that must, as such, affect the victim's reserved domain.

B. Intent and Knowledge

The question whether coercion must be intentional is of far greater import when it comes to coercion-as-control than regarding coercion-as-extortion. The same goes for any requirement of knowledge on the part of the victim state. As I have explained, in coercion-as-extortion the coercing state invariably intends its actions (demand-threat-harm) to produce a change in the behavior of the coerced state, while the latter will invariably know that it is being coerced.²⁴⁷ Whether proving these subjective elements is legally required or not matters little, because proving them will generally be straightforward. But the same is not true of coercion-as-control; the scope of actions that the non-intervention rule would prohibit under this model will depend greatly on how any subjective elements are calibrated.

Consider the following scenarios. State A launches a cyber operation against State B to implant malware into its military command infrastructure. However, the malware also infects the servers of B's election commission and causes substantial disruption to an election taking place in B. In this scenario A did intend to launch the cyber operation in question; in that sense its action was intentional under some minimal description. But A did not act with the intention of disrupting B's election, even though its operation did have this effect. Did A coerce B, i.e., did A intervene in B's internal affairs by disrupting its elections, even though it did not intend to do so? Or, imagine if State A hacked the elections in State B and intended to do so, but the virus it used spreads uncontrollably and also disrupts the elections in State C. Did A coerce not only B, but also C?

On one view, as cogently expressed by the government of New Zealand, coercion requires intention, and with coercion-as-control this would be “an intention to deprive the target state of control over matters falling within the scope of its inherently sovereign functions.”²⁴⁸ From a fair-labeling perspective, intention to deprive the victim state of control over its reserved domain would be crucial for delineating intervention from other possible violations of international law, for capturing what is distinctively wrongful about intervention.²⁴⁹ This is not to say that coercion would require a particular motive, but that the specific consequence of the coercing state's action that would deprive the victim state of control over its reserved domain needs to be intended by the coercing state.²⁵⁰

On the other view, intention to produce a specific effect should not be required for coercion, including coercion-as-control—the effect of depriving the victim state of its ability to control its internal or external should suffice. In the examples above, if State A caused injury to States B and C through its cyber operations (or whatever other activities), it would be normatively undesirable to argue that B and C have no entitlement to reparation from A for the injury caused just because the specific harms were not intended. Moreover, the practical difficulty of proving an intent to deprive the victim state of control over its reserved domain is greater here than with coercion-as-extortion, where inferences of intent are inescapable. While such inferences can reasonably be resorted to in many cases of coercion-as-control, and the difficulties of proving intent should not be exaggerated, they should also not be underestimated. An intent requirement allows the wrongdoing state a plausible way of escaping responsibility by arguing that whatever harm it caused was unintended.²⁵¹

As things stand, it is impossible to provide a definitive answer as to whether coercion-as-control always requires intent or not, pending further evolution in state practice and opinio juris. What can conclusively be said, however, is that without an intent requirement the non-intervention rule would be far broader in scope. Moreover, without such a requirement intervention could, at least in some circumstances, be easier to demonstrate than say violations of sovereignty, even though intervention is generally perceived to carry with it an elevated degree of stigma. Arguing for an intent requirement, as some states have explicitly done, would go a long way to mitigating concerns that the non-intervention rule would be overinclusive.²⁵² Consider, for example, a scenario in which State A emits pollution onto the territory of State B, causing B to take various mitigating measures to protect the environment and its population. In terms of the effect alone, one could reasonably say that A interfered with B's internal affairs, causing it to take actions that it would otherwise not have taken. But absent intent I would personally find it difficult to label such a scenario as coercion, even if there is a clear causal relationship between the actions of the two states.²⁵³

In my view it can also be conclusively said that the subjective elements of coercion-as-control should not include a requirement that the victim state knows that it is being coerced. First, it is generally unusual, and normatively difficult to justify, to require subjective elements from the victim of a wrongdoing, when the wrongdoer is already acting with fault. Second, the intuition that the victim of coercion needs to know that they are being coerced stems from the extortion model, in which coercion works through the exertion of pressure on the victim's will through threatened or inflicted harm, of which the victim will invariably be aware. But that intuition is simply misleading when it comes to coercion-as-control.²⁵⁴

Consider, again, the example of cyber operations interfering with elections infrastructure and machinery, on which so many states seem to agree as the paradigmatic example of coercive intervention. Such operations will generally work (at least to their fullest) only if the victim state is unaware that they are being conducted. Were a state to know that its voting or counting systems were compromised it would simply annul the elections and hold them again, or take some other remedial action. It makes no sense to regard the most successful (and therefore most harmful) such operations—those that go undetected—as failing to meet the requirements of intervention. Moreover, the victim state may become aware of such operations at a later date, rather than at the moment of their execution. It again makes no sense to argue that such operations were not intervention at the moment they were being conducted, but that they became intervention weeks, months or years after the fact, once the victim had become aware.

C. Threshold of Coercion

Calibrating the threshold of coercion-as-control is essential for preventing the overinclusiveness of the non-intervention rule. Requiring intention would to some extent mitigate such concerns. But the fact remains that if a control model of coercion is used, the prohibition of intervention can easily become very sweeping.

The first threshold issue is whether coercion-as-control should cover attempted or incomplete coercive acts. Recall that the answer to that question was positive for coercion-as-extortion—it would make little sense to say, for example, that the United States has not intervened against Cuba through economic means simply because Cuba has managed to resist U.S. pressure.²⁵⁵ But coercion-as-control operates differently, and the same answer is not necessarily required. Consider, in that regard, the following scenario: State A launches a cyber operation against State B to interfere with B's elections. However, B's intelligence and cyber security agencies discover and neutralize this operation before it had any effects. In this scenario A did act, with the requisite degree of intention, using a tool that could have produced the intended result. But it was ultimately unsuccessful and B did not, in fact, lose its ability to control its reserved domain.

To my mind, coercion-as-control should cover such failed attempts.²⁵⁶ Imagine if instead of the cyber scenario State A fomented a coup in State B, inciting and even paying B's high-ranking military officers to overthrow the civilian government—but the coup plot was ultimately exposed and failed. This scenario of the fomenting of a failed coup must be intervention; it simply makes no sense for A to escape liability simply because B managed to resist. The cyber scenario, and any other analogous attempt of coercion-as-control scenario, should therefore be treated in the same way.²⁵⁷ That said, the total or partial failure of any coercive action would be taken into account in the duty of the responsible state to provide reparation, because it affects the nature and extent of the injury suffered.²⁵⁸

When it comes to setting the threshold of a completed action that deprives the victim state of its ability to control its reserved domain, there are several possible approaches, with many of the same considerations at play as with coercion-as-extortion. First, actions that are characterized as uses of force would, in principle, be coercive, at least if carried out with the necessary intention. Second, instances of coercion-as-control could be determined on a case-by-case basis, if there is general agreement among states on the specific type of action in question (as with supporting rebels, fomenting coups, premature recognition, or arguably some forms of cyber election interference).²⁵⁹ But such a casuistic approach is difficult to defend for the reasons I have already given. Third, actions intended to deprive the victim state of ability to control its reserved domain would be coercive if they violated some other rule of international law. But this approach would make it very difficult—and impossible if an intention requirement was rejected—to distinguish violations of the prohibition of intervention from violations of other rules,²⁶⁰ even though intervention carries with it an elevated stigma.

Fifth, and finally, coercion-as-control could require a degree of severity in terms of the impacts of the action on the victim state and its ability to control its reserved domain. For example, a cyber operation that consists of hacking a single voting machine is clearly less severe than one that substantially altered voting tallies or suppressed turnout in key areas. To be clear, the wrong with such an operation is not in that somehow, after a successful plot, the leadership of the victim state will change and will make decisions that the intervening state prefers. The wrong is in interfering in the first place. But clearly some forms of interference are (for various reasons) regarded as worse or more severe than others. This is why it will be easier for states to agree that large-scale interference with election infrastructure is coercive, than for them to agree on the coerciveness of spreading disinformation, whose impacts are more difficult to determine. Similarly, it will be easier for states to agree that a coercion-as-control threshold has been crossed when a states uses cyber operations to disrupt the victim state's response to a pandemic (e.g., cripples testing or vaccine distribution databases and systems), than with regard to low-level infiltrations of the health system, even if they require some remedial action.²⁶¹ And it is again crucial to note that in their official positions, some states have already clearly relied on criteria such as scale or severity to define the coercion threshold.²⁶²

Having said that, while a severity threshold for coercion-as-control is workable in principle, it inevitably leads to gray areas. For example, if we accept that if State A funded rebels on the territory of State B this action would be coercive in character, it is not immediately apparent whether A funding a political party,²⁶³ a non-governmental or religious organization, or a media outlet in B would also be coercive,²⁶⁴ or indeed whether the (lack of) coerciveness should depend on whether the funding is being provided overtly or covertly.²⁶⁵ Or, consider the difficult case of the dissemination of disinformation intended to disrupt public confidence in the country's political system, which can undeniably be effective, but where the magnitude of that effect is difficult to measure reliably.²⁶⁶ At least in principle, however, a disinformation campaign, cyber or otherwise, can severely impact on the victim state's ability to control its internal affairs, namely its political system.²⁶⁷

Similarly, another difficult and particularly cogent case is that of economic measures such as export controls. Consider, for example, the various prohibitions that the United States has implemented on the export of newer microchips to China, with the express purpose of hampering China's ability to develop and produce such microchips. China has labeled these export controls as unlawful under applicable WTO rules—a point on which I express no view.²⁶⁸ Assuming arguendo that the measures are not contrary to international trade law, would they nonetheless constitute coercion-as-control, and thus intervention, in China's internal affairs? Developing microchip production is arguably within a state's reserved domain, and the U.S. measures are certainly having a substantial impact on China's ability to develop such production. Is, therefore, this a case of intervention, or do such export controls fall below some minimum level of severity required for coercion-as-control?²⁶⁹

In short, a severity approach to coercion-as-control can lead to substantially less certainty than the same approach to coercion-as-extortion. That said, it has been clear at least since Nicaragua that some actions can be coercive and thus constitute intervention even absent the demand-threat-harm dynamic of coercion-as-extortion. In my view, a control model of coercion best explains such cases. Determining the precise contours of that model will require more granular jurisprudence, state practice and opinio juris. Moreover, insisting on intentionality and severity as parts of the coercion-as-control threshold would assist in distinguishing intervention from violations of other rules of international law, and justify the elevated stigma that a breach of this prohibition carries.²⁷⁰

VI. Conclusion

My argument in this Article is that coercion, as a constituent element of prohibited intervention, can be understood in two complementary but distinct ways: coercion-as-extortion and coercion-as-control. While similar in their essence and purpose, the two models of coercion operate differently. Conflating them can lead, and has led, to much confusion. Developments in state practice and opinio juris, especially in the cyber context, have enabled us to overcome that confusion. Yet I am not arguing that the two models of coercion are some kind of cyber lex specialis. They are equally valid in the cyber context and outside it, as part of a general rule of non-intervention. The two models I have proposed provide us with an approach that is internally coherent, even if liminal cases remain uncertain and the coercion threshold requires further clarification through state practice. It is fortuitous in that regard that, even outside the cyber context, states have been expressing their views on the lawfulness of coercion, with even Western states, which have traditionally opposed prohibiting economic coercion, now acknowledging that in some cases such measures may violate international law.²⁷¹

I have also argued that state views and scholarly positions on coercion have been shaped structurally by the problem of justification. While intuitively we regard some forms of coercion to be justified, the non-intervention rule is framed categorically, as admitting no exceptions. It has no inbuilt balancing test that could accommodate pressuring other states for good reasons. The prohibition of intervention is a rule that is, by its design, structurally apt at being either overinclusive or underinclusive, thus incentivizing arguments about its scope.

In particular, the tendency of states and some scholars to argue that some forms of state actions, such as economic pressure, are inherently not coercive, is directly shaped by their view that sometimes such measures can be used for good ends. But because they cannot argue that coercion is justified, they are compelled to argue that there is no coercion at all, especially when the measures concerned are otherwise lawful. This is in my view deeply problematic, and inevitably leads to a distorted, moralized understanding of what coercion is or is not.

A more promising approach to the issue of justification—perhaps an imperfect one, but still an improvement—is to argue that undoubtedly coercive measures directed to compelling compliance with existing legal obligations, regarding which the target state enjoys no measure of discretion, do not constitute intervention at all. This is because such measures do not interfere with the target state's internal or external affairs on matters on which international law gives it free choice.²⁷² Such coercion does not constitute intervention, although it may violate other rules of international law.²⁷³ While this understanding of the reserved domain element enables unilateralism to some extent, it does not do so unduly. In these cases it is not for the prohibition of intervention, but for other rules of international law, such as sovereignty, human rights, or WTO law, to regulate any harmful effects of coercion.

I have also examined several possible approaches to the threshold question regarding both coercion-as-extortion and coercion-as-control. One alternative to underinclusive minimalism is to equate coercion with the violation (or threatened violation) of the rights of the target state and its nationals. This would cover cases such as economic measures taken in violation of international trade law or investment law, or those of “hostage diplomacy” that violate the human rights of the state's nationals. Or, we could conceptualize coercion as the threat or infliction of harms (in coercion-as-extortion), or the performance of certain actions that materially deprive the victim state of its ability to control its internal or external affairs (in coercion-as-control), that reach a certain level of severity in terms of their potential or actual impacts. There is much to commend in such an approach, but its inevitable consequence is a higher degree of indeterminacy.

One possible response to my overarching argument would be that coercion should not be regarded as a constituent element of prohibited intervention at all, partly because it may be impossible to come up with a principled approach to defining it.²⁷⁴ Pragmatically that seems an unlikely direction of travel. States recently coming out with their formal positions on the application of the prohibition of intervention to cyberspace have all mentioned coercion as an essential element of intervention, often relying on the ICJ's Nicaragua judgment as an authoritative restatement of that principle.²⁷⁵ Not a single state has argued in favor of abandoning the coercion requirement, even in some limited category of cases. Discarding coercion for some superior alternative (even if one existed) therefore does not seem feasible, although such an option may become more plausible if states start endorsing it.

As a matter of policy, I would suggest that, in expressing their opinio juris, states could usefully address three sets of issues in any future statements on the non-intervention rule. First, they should clearly distinguish between coercion-as-extortion and coercion-as-control, while affirming the validity and complementarity of the two models of coercion. Second, they should clarify their approach to the threshold of harm in either model. Third, they should also clarify the nature and role of intention in defining coercion. In that regard, while statements mentioning specific examples of prohibited intervention, such as different forms of election interference, are very useful, it would be even more useful for states to elaborate on why any particular examples are regarded as constituting coercion. I have explained what the plausible options on these issues questions are, but it is ultimately only states that can clarify the non-intervention rule further, and today there are ample opportunities to do so.