Framing Positive Obligations under the European Convention on Human Rights Law: Mediating between the Abstract and the Concrete

Abstract

Positive obligations under the European Convention on Human Rights can be framed with different levels of concreteness. The level chosen is essential for understanding the analytical distinction between the existence of an obligation and its breach. The level of concreteness is an important conceptual framework because it has an impact even on the possibility of making an assessment as to whether the State has breached the obligation, and on how this assessment is performed in the reasoning. Kurt v Austria is used to illustrate how positive obligations can be framed both in more abstract and concrete terms, and how the reasoning mediates between the two. The more it tilts towards a concrete formulation of the obligation, the more the Court appears to assume the role of a rule-maker, which is in tension with the principle that States have discretion as to the concrete measures to fulfill their positive obligations.

1. INTRODUCTION

The development of positive obligations has been one of the hallmarks of the work of the European Court of Human Rights (ECtHR or the Court) in interpreting the European Convention on Human Rights (ECHR).¹ These obligations have been predominantly examined in light of the specific subject matter (e.g. domestic violence, environmental pollution and natural hazards). They have also been systematized in various ways (e.g. substantive v. procedural obligations, explicit v. implied procedural obligations and obligations with preventive functions v. with remedial functions).² These systematisations are useful for better understanding the content of positive obligations (i.e. the measures States need to undertake) and the circumstances when they can be found to be breached. This article takes a novel approach in that it proposes an examination of positive obligations from the perspective of the level of concreteness chosen for their framing. This level refers to the choice of how specific the framing is, given the particular facts.³ I argue that this level is crucial for grasping the nature of positive obligations and how they can be breached. More specifically, the choice made as to how they are framed is essential for understanding whether and how the analytical distinction between the existence of an obligation and the breach of this obligation can be applied to them. The level of concreteness chosen for articulating positive obligations is important because it has an impact even on the possibility of making an assessment whether the State has failed to fulfil the obligation and on how this assessment is to be performed in the legal reasoning.

To explain the importance of the level of concreteness used in the framing of positive obligations, the article proceeds at follows. Section 2 explains the importance of the analytical distinction between determining the existence of an obligation, on the one hand, and the breach of this obligation, on the other. This distinction poses challenges when an omission is the basis for state responsibility. Although similar difficulties can be identified in the determination of the breach of negative obligations, the challenges with omissions are still distinctive. As Section 2 clarifies, the reason is that the review of the breach of a positive obligation that implies the assessment of omissions includes a wider scope of alternatives as possible counterparts to the claimed omissions. Having identified the necessary steps and inevitable challenges in the legal decision-making when an omission is framed as the basis for state responsibility, Section 3 proceeds to explain the steps actually followed by the ECtHR as a decision-maker. It identifies three levels of concreteness in how the ECtHR has chosen to frame positive obligations. These levels reflect a progressive specification in that Levels 1 and 2 are more abstract and thus less concrete. No determination of breach is performed at these levels; rather, the Court aims to articulate some general and abstract standards. Level three is the most concrete since it entails the application of these standards to the specific factual circumstances and identification of specific omissions.⁴ The concrete positive obligations that can be derived from these standards are, however, not articulated prior to the determination of breach. This implies that no analytical distinction is made between the post factum determination of the obligation (that is assumed to have existed ex ante, i.e. at the time when the events were unfolding) and its breach; rather, the whole analysis at Level 3 collapses into the post factum determination of breach. This is problematic since, as Section 2 notes, the establishment of State responsibility is based on the assumption that an obligation first exists for the State and only then it can be actually breached. Yet, this assumption that implies a demarcation between the existence of an obligation, on the one hand, and the breach of this obligation, on the other, disregards that the obligation can be farmed with different levels of specificities, as the actual reasoning of the Court demonstrates. When this specificity reaches its most concrete level (i.e. what I have framed as Level 3), the collapse that will be concretely demonstrated in this article, is analytically inevitable.

The collapse reveals the nature of positive obligations and the role of the Court in adjudicating them. Specifically, as I will show, positive obligations are underpinned by two dominant and competing ideas working under the surface of concrete court cases: the formulation of rules for States (i.e. the rule-making role of the Court) versus the application of rules to facts (i.e. the rule-application role of the Court). The second role necessarily requires some specifications of the positive obligations, and different choices can be available on how to specify them. Making these choices can imply that the Court engages in rule-making, undermining States’ discretion to choose the measures for fulfilling positive obligations. Admittedly, when the Court adjudicates negative obligations, it adopts similar roles.⁵ Yet, there is still a nuance since, as Section 2 explains, the scope of choices is wider in the context of positive obligations given the wider scope of alternatives as counterparts to the claimed omissions. In addition, by specifying concrete alternative measures that should have been undertaken, as opposed to identifying conduct that the State should have abstained from to fulfil its negative obligations, it is easier to perceive the Court as adopting the role of a rule-maker.

It has to be also acknowledged that the Court adopts the role of a rule-maker at all of the three levels of concreteness used in the framing of positive obligations (i.e. Levels one, two and three). Even at Levels 1 and 2, standards are formulated. However, since these formulations are abstract and no determination of breach is performed (see Sections 3.A and 3.B), it is only at Level 3 where the rule-making role can be considered as most controversial. The reason is that at Level 3, concrete measures need to be formulated for assessing breach, which can be perceived as being in tension with States’ discretion to choose the concrete measures for fulfilling positive obligations. As Section 2 shows, this discretion is an important starting point in the reasoning regarding breach of positive obligations, which adds a distinguishing nuance when compared with negative obligations. ⁶ As I will further explain in Section 3, by invoking the standard of reasonableness, the Court aims to preserve States’ discretion. Yet, the assessment whether the undertaking of certain measures is reasonable necessarily requires the formulation and consideration of alternative measures that need to be specified. This formulation and specification of the alternatives, as inevitable steps in the legal reasoning, renders the above-mentioned tension also difficult to avoid.

A note on the methodology is due. The law on state responsibility is an important starting point for understanding the distinction between the existence of an obligation and its breach.⁷ The law on state responsibility also provides certain tools, such as the classification of obligations into obligations of means, and of result, tools that are invoked for facilitating the reasoning in finding a breach.⁸ The method that views the ECHR as included in general international law calls for the following clarifications. Although the Court develops its own jurisprudence for the protection of human rights⁹ and is influenced by the national legal systems,¹⁰ as an international treaty, the ECHR is also embedded in general international law.¹¹ In this sense, the violation of the ECHR necessary implies the establishment of state responsibility under international law. It is therefore not unconventional to analyse the ECHR from the perspective of state responsibility as developed in general international law.¹² Although such an approach acknowledges the distinctiveness of human rights law, it still considers it as integrated, rather than isolated, from other areas of international law.¹³ Such an approach is in fact followed by the Court itself. The Court has developed an interpretative principle to take other international law norms into account.¹⁴ It has also clarified that it determines state responsibility ‘in conformity with the governing principles of international law, although it must remain mindful of the Convention’s special character as a human rights treaty’.¹⁵ This integration is not only accepted by the Court but also by important actors mandated to interpret and develop general international law, including state responsibility.¹⁶

Although responsibility in human rights law is part of the law on state responsibility,¹⁷ the latter does not suffice for an in-depth analysis of the primary rules the Convention is part of and the positive obligations it imposes.¹⁸ These obligations cannot be understood without a serious engagement with the ECtHR case law. This poses a challenge since the case law is very rich. Yet, the Court has developed certain analytical steps in its reasoning that are normally repeated. The Grand Chamber (GC) judgements are also considered to have an important value.¹⁹ For this reason, Section 3 focuses on Kurt v Austria delivered by the GC in 2021 to illustrate the different levels of concreteness in how the Court frames positive obligations.²⁰ The analytical steps that the Court normally applies to positive obligations in terms of first formulating some general standards (as reflected in what I frame as Levels one and two) and then their specification (Level three), are repeated in Kurt v Austria and in this sense, the reasoning in this judgement is a continuation of the Court’s argumentative methodology. Since it is not likely that the Court will change this methodology, the analysis offered in this article with focus on Kurt v Austria can be expected to have a general validity.

There is an additional crucial reason as to why I focus on a specific case. As I explain in Section 2, the problem regarding the destabilization of the distinction between the existence of an obligation and its breach can be framed as a problem of specification of obligations. To explain precisely this specification and the different levels at which it has been done, a specific case is necessary. Level 3 of specification where the determination of breach is actually done, cannot be elucidated without focusing on a specific case and without a detailed analysis of the reasoning in this very case. Therefore, the best methodological approach that enables me to explain Level 3 of specification is focusing on a specific case. It is only in a specific case, where the question of breach arises, where concretely the scope and the content of the concrete positive obligation (i.e. Level 3) are formulated, and where choices as to the concretisation and the formulation have to be made. In other words, the understanding of the analytical steps actually followed for establishing breach requires a close scrutiny of the reasoning in a specific judgement.

Certainly, this implies certain limitations. Perhaps the most important one is that Kurt v Austria is a domestic violence case and an objection can be anticipated that, as such, it cannot be representative of all positive obligations cases. This is an objection that will be valid for any judgement that might be selected. Yet, as explained in the previous paragraph, a selection is imperative for understanding how positive obligations are specified and for making the analysis of specification at Level 3 manageable. In this sense, the analysis sought to be performed makes the chosen method (i.e. focus on a specific case) indispensable. It is also vital to note that my analysis in Section 3 is sensitive to the specificities of the domestic violence context as one of the many contexts where positive obligations have been developed by the Court. Most importantly, however, no claim of general representativeness of the entire category of positive obligations is attempted. This article rather seeks to illustrate how specification has been performed by the Court and what analytical challenges arise in performing it. The analysis applied to Kurt v Austria, particularly in terms of identifying the challenges in the specification, could be applied by analogy to any other ECtHR judgement addressing positive obligations. In this specific sense and despite the methodological limitations, the analysis still has general theoretical relevance.

One can still ask why my qualitative analysis that aims to identify the challenges in the specification of positive obligations, focuses specifically on Kurt v Austria. Certainly, as already mentioned above, being a GC judgement where the Court in great detail systematizes, reiterates, develops and tries to justify its reasoning on positive obligations is key. The different levels of specification of the positive obligations are thus very clear in the separate parts of the judgement. Critically, all substantive positive obligations were at stake in this case. An additional consideration for my choice was that Kurt v Austria can be viewed as a strategic critical case, where the applicants suffered very serious harm and, yet, Austria was not found to have failed to fulfil any specific positive obligations (i.e. Level three). Despite this negative finding, as this article will show, by using the different levels of specification of the positive obligations, the Court still tries to develop protective standards.²¹ Furthermore, the finding of no violation at Level 3 was far from uncontroversial given that seven judges dissented, which also confirms the characterisation of Kurt v Austria as a critical case that offers important insights of strategic importance.²²

As much importantly, I have resorted to tools to mitigate the methodological limitations that necessary follow from my focus on a specific case. First, where relevant, references to other important judgements issued prior to Kurt v Austria are also provided to track the continuation and any new nuances. Second, specific engagement with the case law after Kurt v Austria is also important to track whether and how the argumentative choices have been repeated in later judgements. Although not all of them are cited, I have reviewed in the HUDOC database all judgements issued after 15 June 2021 when the GC delivered Kurt v Austria till 7 November 2022, involving substantive positive obligations under Articles 2 and 3 ECHR, while judgements where the only claimed omission was failure by the State to investigate or to have effective judicial response are excluded.²³ Here, it is relevant to note that the Court applies similar approaches to its review of positive obligations under Articles 2 and 3²⁴ that have been framed as reflecting ‘basic values’.²⁵ Although Kurt v Austria was reviewed only under Article 2, the similarity in the approach to Articles 2 and 3 justifies the extension of my search in the HUDOC database to both provisions to track any subsequent developments. This extension also helped me to track any nuances in the specification of positive obligations in situations of non-deadly violence.²⁶

The forthcoming analysis excludes positive obligations delivered under qualified rights, such as Article 8. This exclusion is justified given the different structure of these rights, including the uncertainty whether they trigger positive obligations as a matter of principle under all circumstances where the definitional scope of Article 8 is engaged.²⁷ As to the exclusion of judgements whose focus is on the obligation to investigate or to have an effective judicial response,²⁸ this obligation was not reviewed in Kurt v Austria. The exclusion of the procedural limbs of Articles 2 and 3 is without prejudice to the potential value in analysing them from the perspective of the different choices for specifying investigative measures that might or could have been undertaken by the national authorities.

In sum, although the applied method suffers from limitations and although the Court might depart from some aspects in the reasoning applied in Kurt v Austria, a departure that given my analysis in Section 3 might be desirable in some respects, this does not invalidate the theoretical value of the forthcoming analysis that is structured around a specific case. On this ground, Section 3 follows the steps that characterise the reasoning in Kurt v Austria.

2. THE DISTINCTION BETWEEN THE EXISTENCE OF A POSITIVE OBLIGATION AND ITS BREACH

Positive obligations become relevant when an argument is made that a State has omitted to do something. Because multiple omissions might be attributable to the State, the application of an analytical tool is needed to delimit those omissions that are legally relevant. In the law of state responsibility, this tool is the distinction between obligation and breach, which implies that it first needs to be determined that the State was under an obligation to act (i.e. not to commit an omission) before an assessment as to the breach of this obligation can be performed. As this section will show, however, this distinction is difficult to apply in the assessment of state responsibility for the breach of positive obligations. Its disruption is actually inevitable since the distinction does not reflect that the decision whether a positive obligation has been breached necessarily demands the specification of alternative measures. This specification can only be done in light of the concrete facts (what I frame as Level three specification) and thus unavoidably post factum.

A. Omissions and their Multiple Alternatives

Article 2 of the International Law Commission (ILC) Draft Articles on Responsibility of States for Internationally Wrongful Acts stipulates that ‘[t] here is an internationally wrongful act of a State when conduct consisting of an action or omission: (a) is attributable to the State under international law; and (b) constitutes a breach of an international obligation’. The Commentary to the ILC Draft Articles clarifies that ‘[…] no difference in principle exists between the two [an action and an omission]’.²⁹ On closer analysis, to say that there is no difference between actions and omissions is to seriously understate the complexity of establishing responsibility for omissions. Admittedly, the distinction between acts and omissions to collective agents and social institutions, such as the State, might be difficult to make, if we do not have ‘baseline comparisons’³⁰ and certain normative positions that can guide us.³¹

Yet, an omission cannot cause harm in the same way as an action. Moral philosophy has engaged with this issue and made the point that causing harm by action cannot be equated with causing harm by omission.³² The national constitutional law,³³ administrative law and tort law also have adopted the distinction and developed complex analytical frameworks for examining when state authorities can be held responsible for omissions.³⁴ The issue of causation has played a major role within these frameworks, that is, the causal connection between the omission and the harm.³⁵ However, causation is far from being the single criterion; policy and other considerations (e.g. knowledge and foreseeability of the harm, the reasonableness of imposing a positive duty to act) also intervene in the assessment,³⁶ clearly demonstrating that establishing responsibility for omissions is an exercise fraught with complications.

Besides moral philosophy and national law, the ECtHR also insists on the distinction between,³⁷ on the one hand, negative obligations breached because of an action by the State and, on the other hand, positive obligations breached due to omissions. Surely, in the context of Article 8, the Court often refuses to specify whether it will examine the case as one implicating positive or negative obligations.³⁸ This refusal does not necessary reflect the Court’s conception of obligations but rather its methodology. As Vorland Wibye has explained rather than getting involved in theoretical and conceptual discussions about the distinction, ‘the more efficient approach when tackling individual applications is to proceed directly to proportionality review and the ultimate determination of whether the state has struck the right balance between active measures and non-interference’.³⁹

Despite the inconsistency, the ECtHR still relies on the distinction. Yet, the delineation criteria have not been clearly articulated in the case law. This failure can be related to the refusal by the Court to ‘to develop a general theory of the positive obligations which may flow from the Convention’.⁴⁰ Neither the failure nor the refusal is surprising. The Court does not engage with complex philosophical discourses, and, in any case, the problem of distinguishing causing harm by action versus by omission is persistent.⁴¹ Although in some situations, the distinction is easier to discern (i.e. like in domestic violence cases where a private actor causes harm and there is no action by the state that has directly caused this harm), there is no normatively neutral way for advancing one single general justification of the distinction in all possible situations.⁴² It is beyond the scope of this article to engage with such a general justification.

What suffices to note here is that besides the difference between actions and omissions that can serve as a relevant (although, as already observed problematic and thus not all the time sufficient) reference, another related basis for understanding the distinction is that the means (i.e. the concrete measures) for complying with a positive obligation are multiple. In particular, Alexy has usefully explained that there is a structural distinction between defensive rights (that have negative obligations as corresponding obligations) and the rights to positive state action: ‘if there is a command to protect or support something, then not every act which represents or brings about protection or support is required’.⁴³ There is thus a variety of ways to protect and different means for achieving certain ends. Alexy observes that if there is one single suitable protective or supportive act necessary to satisfy the protective right, then the structure of the protective right will match that of defensive rights. If there is only one effective means, the state must adopt it, which appears to reflect straightforward situations. However, beyond these situations, positive rights have alternative and disjunctive structure. As opposed to the conjunctive structure of defensive rights, the alternative structure of positive rights means that an omission has no definitive counterpart. It can be argued, however, that there is a range of reasonable alternatives and it might be possible to advance a range of actions to protect the rights.⁴⁴

It follows then that the availability of multiple options/measures for complying with positive obligations increases the scope of state discretion and makes the demonstration of breach more difficult. This is clearly reflected in the case law of the Court, where a crucial assumption is that States have a choice how to comply with positive obligations,⁴⁵ and there is a ‘wide range of possible measures’ that they can undertake.⁴⁶ Importantly, not every possible measure that brings about compliance with the obligation is required, since there are ‘multiple, independently sufficient paths to compliance’.⁴⁷

At this juncture, it needs to be also acknowledged that States can also use different measures for limiting rights, some of them possibly in breach of negative obligations. This means that the availability of multiple options and alternatives is also necessarily present as an analytical step in the review of the breach of negative obligations. This also means that the issues that I will discuss below (e.g. the distinction between the formulation of the content of the obligation and the establishment of breach, the existence of different alternative measures for limiting rights and the application of the standard of reasonableness) are also relevant to negative obligations.⁴⁸ There is, however, still a difference. In particular, all measures that constitute disproportionate limitations are in breach of negative obligations.⁴⁹ As Vorland Wibye has noted, once a limitation measure passes the threshold of disproportionality, the only way to comply is to abstain from this measure.⁵⁰ One can rebut that States still have a wide range of possible measures at their disposal how to limit rights. However, the normative starting point is that States have to choose the least restrictive measures when they take actions to limit rights.⁵¹ In this way, their choice of measures is more circumscribed. In contrast, the Court has never formulated a test to the effect that States have to undertake the most protective measures to ensure the rights.⁵² The starting point is rather that States can choose the measures and their failure to choose the best measure for protecting a person (arguably in fulfilment of a positive obligation) does not necessary lead to a breach. In comparison, when the Court adjudicates negative obligations, its starting point is not that States have different means of restrictive rights and that even if one restrictive measure is disproportionate, the proportionality of other measures will still be examined. If one measure limiting the right is disproportionate, this measure is straightforwardly in breach of negative obligations.⁵³

There is therefore a similarity between positive and negative obligations in that the review of both includes consideration of alternatives. The similarly, however, does not negate the validity of the forthcoming analysis. The subsequent analysis in this article is not premised on the demonstration of the categorical validity of the distinction in all possible situations. In fact, similar analysis regarding, for example, the importance of alternatives for limiting rights and the different levels of concreteness for formulating such alternatives, could be performed for negative obligations, which could be an object of a separate article.

Neither does the existence of the above similarity negate more generally the distinction between positive and negative obligations. The wider choice of means/alternatives that characterise compliance with positive obligations and that can be used as a reference point for the distinction between positive and negative obligations makes the finding of breach more difficult because of the need to consider wider alternatives and counterfactuals (i.e. what other means could be have been used to ensure the right). As Section 3 will show, these alternative means of protection can be framed in different ways and with various levels of specifications, which intensifies the difficulty.

Neither does the existence of similarities invalidate the relevance of the distinction between acts and omissions. Alternatives to omissions are more problematic than alternatives to actions that are relevant to consider in the least restrictive mean test for reviewing the breach of negative obligations. Alternatives to omissions are more problematic for at least two reasons. First, omissions themselves, as opposed to actions that limit rights, are more difficult to perceive and therefore to specify. Second and relatedly, alternatives to these already perceived omissions are also more difficult to specify in a definitive way. As Klatt has observed, ‘unlawful omission of an action has no definite opposite’.⁵⁴ I have to accept that there is some level of circularity here since the absence of definitive opposites, which implies the availability of multiple options for fulfilment of positive obligations, is premised on the distinction between acts and omissions. This suggests that it is ultimately difficult to completely separate our conceptualization of obligations from the distinction between acts and omissions. In sum, basing state responsibility on omissions is problematic because omissions introduce ‘into judicial reasoning a purely hypothetical factor’.⁵⁵ They imply a counterfactual analysis. In addition, there are numerous omissions that can be attributed to the State. Many of these might not even be cognizable, at least not prior to the actual materialization of harm. It would be absurd to suggest that every single omission that can be perceived/identified and that can be somehow related to the harm suffered by an individual (or the risk of harm) should be a basis for establishing state responsibility. There need to be ways of establishing the boundaries of the legally relevant omissions, which implies certain analytical tools that guide the assessment of responsibility.

One such fundamental analytical tool is the distinction between the existence of an obligation and the breach of that obligation. In simple words, there must be an obligation to act in the first place that has remained unperformed, against which state responsibility can be assessed. As Crawford has framed it,

[…] omission is more than simple ‘not-doing’ or inaction: it is legally significant only when there is a legal duty to act which is not fulfilled, and its significance can only be assessed by reference to the content of that duty. […] the absence of any primary obligation ‘to do’ will mean that no omission may be complained of.⁵⁶

The issue of timing seems to be important here, since it implies a distinction between an ex ante existence of an obligation to act, and the ex post assessment whether this obligation (i.e. the duty to act) has been performed. The challenge with omissions and positive obligations is framing the obligation to act ex ante, which means prior to the time when the events unfolded.⁵⁷ The difficulty in framing the obligation ex ante implies that the content of the obligation (the measures that the State should have taken) and the scope of the obligation (how many measures or how intrusive these measures should be) are difficult to frame.⁵⁸ Not only is the ex-ante framing of the obligation and its content and scope a challenge; it might even be debatable whether there was an obligation at the time when the events were unfolding.⁵⁹ The answer to this question depends on whether and how an omission can be framed. In addition, and assuming that there was an obligation, the latter can be framed at different levels of abstraction and with different levels of concreteness.

B. Post Factum Assessment of Omissions and the Importance of their Specification

Articulating the positive obligation ex ante, so that the state authorities know how to react to risks, and articulating these obligations post factum, so that the ECHR can assess compliance, is a challenge because the ECHR (and more generally human rights law treaties) uses rights as its organizing principle.⁶⁰ Linking these rights to corresponding positive obligations is conceptually complicated for at least two reasons. First, as mentioned above, it is indeed a challenge to frame these obligations ex ante. Yet, it is not desirable to categorically remedy this problem since human rights law is meant to be dynamic by generating plurality of duties, individual-centred and thus context dependent.⁶¹ Second, the assessment of breach is done with reference to variables such as causation and reasonableness, which are not easy to apply.⁶² The Court has to make an assessment without the benefit of hindsight,⁶³ which means it has to decide whether at the point in time when the State should have arguably acted (arguably in fulfilment of its positive obligation), it was actually reasonable to act and the act could have contributed to preventing the harm.

It then follows that despite the affirmation by the Court that the breach of positive obligations is assessed without the wisdom of hindsight, one can doubt whether this is actually possible for at least two reasons. First, the assessment would not be done if harm had not materialized (i.e. if there were no victim).⁶⁴ Importantly, the assessment as to the existence of an obligation—including its content and scope—and its breach, can be contingent on the gravity of the harm, and this gravity becomes known only post factum (i.e. after the State arguably breached its obligation and an assessment as to this breach needs to be made).⁶⁵ The severity of the harm might affect the scope and content positive obligations,⁶⁶ and in some contexts, the Court has been explicit to this effect.⁶⁷ At the time when the events are unfolding, the final and actual harm is not known. Once it becomes known (i.e. when it materializes), the easier it is to make an argument that the national authorities ought to have known about it.

Second, it is doubtful whether it is possible to apply the test of ‘ought to have known’ about the risk of harm (a test that the Court has developed for deciding whether a State was under a positive obligation and whether it has breached it)⁶⁸ without some benefit of hindsight. The test of ‘ought to have known’ necessarily implies some retrospective assessment performed by the Court after the materialization of the harm. During this retrospective assessment, the Court has to simulate ignorance. However, as its name suggests, the test of ‘ought to have known’ implies some normative assessment that can be influenced by circumstances knowable and known only post factum.⁶⁹

Even if we accept that it is possible to make the assessment without the benefit of hindsight, the assessment is guided by certain factors. Knowledge, causation and reasonableness were already mentioned. The last deserves special attention since the content and scope of positive obligations can only be within reasonable limits. The State is only required to take reasonable measures.⁷⁰ The factor of reasonableness for the assessment of breach necessarily presupposes a context-specific assessment, and hardly any ex ante determination can be made as to what is reasonable to expect from the State. It might be possible to determine ex ante in abstract terms what is reasonable in situations where there is risk of harm; however, the articulation of the measures that form the positive obligation in more concrete terms in light of the concrete circumstances, might be possible only post factum.

At the point, a return to the distinction between positive and negative obligations might be apt. Although the focus of this article is on the conceptual challenges for determining the breach of positive obligations, it is still relevant to observe that reasonableness and proportionality review for determining breach is always a post factum review. The same is equally valid for negative obligations. However, the review can be expected to be more complex in cases of omissions due to the wider compliance options, as explained in Section 2.A. The lines of causation can be more complex, and the knowledge can be more difficult to establish. In contrast, if the State acts to limit rights, the causal links and thus state knowledge are easier to discern. For example, if a state agent acts, possibly in breach of negative obligations, causation is not an issue since the agent’s actions are directly attributable to the State,⁷¹ and if they are, knowledge of the State is assumed.⁷²

In sum, given the factors of causation, knowledge and reasonableness, it cannot be ascertained with certainty in advance whether there is a positive obligation and what that obligation is, including its content (i.e. what measures/actions should have been undertaken) and scope (how far-reaching and demanding these measures/action should have been). Once harm materializes, an ex post facto assessment can be performed. It is also important to highlight here that these factors (knowledge, causation and reasonableness) are used not only in assessing whether there is a positive obligation and what its content might be but also in the assessment of breach. This facilitates the convergence of the analysis of the two analytical steps, namely, the existence of an obligation and the breach of that obligation.⁷³ More specifically, both analytical processes (making the ex post facto assessment as to whether an obligation existed at the relevant point in time in the past and whether it was breached) demand that the decision maker (i.e. the ECtHR) has some idea as to the concrete reasonable and contributory measures that could have been adopted. Otherwise, it will be difficult to perceive and measure the omission as a basis for responsibility. The decision maker needs to consider what measures should have been taken (if none whatsoever were actually undertaken) or what alternative protective measures could have been undertaken (if some were undertaken but they arguably turned out to be ineffective).⁷⁴ The omission needs to be framed and articulated before the obligation can be framed and articulated. This framing can be done in different ways depending on the level of concreteness and abstraction applied.

Any positive obligations can therefore only be framed post factum, meaning after some harm has materialized and after an actor (most likely the victim of this harm),⁷⁵ articulates an action that the State should have taken, in this way formulating a possible obligation. I say ‘possible’ since this articulation and formulation does not necessary mean that the Court will agree that the State was actually under such an obligation.

The possibility to frame the obligation only post factum questions the distinction clarified in Section 2.B. The actual problem is, however, that although this distinction is fundamental in the law on state responsibility,⁷⁶ including state responsibility under the ECHR,⁷⁷ it is not mindful that the process of making a decision as to whether breach should be found necessary implies specifications of omissions and alternatives in light of the concrete facts (i.e. Level 3). Choices can be made as to which specifications will be selected and how they can be articulated. Importantly, such choices that inevitably have to made so that a determination of breach can be performed, are possible only post factum. The hindsight problem therefore occurs in relation to the delineation of the concrete measures that should have been undertaken, which corresponds to what I frame as Level 3. The hindsight problem does not seem to be poignant at Levels 1 and 2, whereas Sections 3.A and 3.B will show abstract positive obligations to generally ensure that the rights have been formulated ex ante by the Court in its various judgements.

As explained in Section 2.A, the choices of specification (at Level 3) are wider in the context of positive obligations, which can make the determination of breach more difficult and the hindsight problem more poignant, in comparison to negative obligations. It then follows that the problems mentioned above are related to the standards of reasonableness, causation and knowledge that destabilise the analytical distinction between determining the existence of an obligation, on the one hand, and the breach of this obligation, on the other, are actually problems about specifications and choices what specifications to use. To understand therefore the post factum determination of breach of obligations due to omissions, it is important to understand how the Court performs specification in its reasoning.

3. LEVELS OF CONCRETENESS IN FRAMING POSITIVE OBLIGATIONS

To illustrate the levels of concreteness, the GC’s reasoning in Kurt v Austria will be used. Although one needs to be mindful of the methodological limitations that my focus on a single case poses, as already observed in Section 1, the specification of obligations can be best explained precisely via a detailed analysis of a specific case.

The applicant in Kurt v Austria alleged that the Austrian authorities had failed to protect her and her children from her violent husband and that this resulted in him murdering their son. The child was shot at his school. Before the fatal incident, the authorities issued a barring and protection order that obliged the perpetrator to stay away from their common apartment, the applicant’s parents’ apartment and the surrounding areas. The applicant argued that this order was an insufficient measure. Instead, the authorities should have held the man in pre-trial detention. The alternative measure, i.e. the content of the positive obligation as framed by the applicant, therefore, was pre-trial detention. The applicant also invoked other measures that the State failed to fulfil, namely, a barring order regarding the school and the sharing of information with the school authorities about domestic violence.⁷⁸

Prior to examining the analytical steps performed in the reasoning, a brief description of the Court’s findings is due. The Court held that Austria was not under the concrete positive obligation to hold the man in pre-trial detention,⁷⁹ nor was it under the obligation to take the other measures invoked by the applicant. The Court formulated, however, additional measures to assess the conduct of the respondent State; it concluded that these were either fulfilled (e.g. the conduction of a risk assessment regarding the woman) or the State was not under an obligation to perform them (e.g. taking protective operational measures to protect the life of the child). This led to the conclusion that Austria did not fail to fulfil its positive obligations under Article 2 ECHR.

Three analytical steps can be distinguished in the Court’s reasoning to reach these conclusions: first, framing of general positive obligations under Article 2; second, framing of general positive obligations under Article 2 in the specific context of domestic violence; finally, concretisation of these obligations to the specific case. For the sake of simplicity, I will refer to these steps as Level 1 (the most abstract and general level of framing positive obligations), Level 2 (the intermediary level of framing positive obligations) and Level 3 (the most concrete level of framing positive obligations). Below, I will review each one of these steps to understand whether and how the distinction between the existence of an obligation and the breach of this obligation is applied. The review of these steps will also illustrate how the Court reasons when multiple proposals are possible as to the concrete measures that could have been adopted by the State and how the Court navigates between the two roles of rule making and rule application.

A. Level 1—General Positive Obligations under Article 2

The first step in the Court’s reasoning is the most abstract, since it starts its analysis by observing at para 157 that ‘[t]he first sentence of Article 2(1) enjoins the State not only to refrain from the international and unlawful taking of life, but also to take appropriate steps to safeguards the lives of those within its jurisdiction’. The obligation of taking appropriate steps to safeguard lives is formulated at a very high level of abstraction. The framing is result-oriented since the emphasis is on the objective of safeguarding lives. This framing tells us nothing about the steps (i.e. the measures), other than that they have to be appropriate. This initial formulation of this abstract positive obligation, however, suggests that the State as a matter of principle is constantly under the obligation to safeguard lives. This means that every time harm materialises (death or a life-threatening situation)⁸⁰ that might trigger a post factum assessment as to a possible breach of the ECHR, the assumption is that the State was necessarily under the obligation to take appropriate measures to prevent that harm. This assumption resembles the French law on administrative liability,⁸¹ where the State, due to its very nature and purpose, is assumed to be under a general obligation to administer competently within its jurisdiction. The abstract obligation of taking of appropriate steps to safeguard life is separated from the question whether any measures were successful in terms of actually preventing concrete harm.⁸²

This abstract obligation entails the obligation to criminalise, the obligation to develop an effective regulatory framework, the obligation to take preventive operational measures and the obligation to conduct a risk assessment.⁸³ The first has been framed in this way at para 157:

a primary duty of the State to secure the right to life by putting in place effective criminal-law provisions to deter the commission of offences against the person, back up by law-enforcement machinery for the prevention, suppression and punishment of breaches of such provision.

The obligation to criminalise was not relevant in Kurt v Austria, and I will not address it further. It suffices for the purposes of this analysis that the obligation to criminalise is a concretisation of the abstract positive obligation of taking ‘appropriate steps to safeguard’ lives. The obligation to criminalise also does not normally raise challenging analytical issues regarding the application of the above-mentioned factors of causation, knowledge and reasonableness. In its reasoning, the Court assumes that it is reasonable that grave forms of harms are criminalised at national level, which is important inter alia for their prevention.⁸⁴ It is normally the actual implementation of the national criminal law, in terms of effective investigation and quality of the criminal proceedings, that the Court has to engage with in its case law.⁸⁵

Criminal law is only one part of the national regulatory framework. For this reason, the Court has also framed the obligation of developing an effective regulatory framework ‘to afford protection against acts of violence by private individuals in any given case’,⁸⁶ which is also a concretisation. There are multiple national legal frameworks that might have preventive functions. Which of them might be pertinent depends on the factual circumstances of the concrete case and, equally importantly, on which omissions are identified (or identifiable) as causative. This is a point to which I return in Section 3.C.(iv) with reference to Kurt v Austria and the avoidance of the GC to articulate an omission with reference to the national regulatory framework.

Besides criminalisation and an effective regulatory framework, a third positive obligation has been formulated by the Court in the following way: ‘positive obligation on the authorities to take preventive operational measures to protect the individual whose life is at risk from the criminal acts of another individual’.⁸⁷ This is the so-called Osman test, which has been well established in the case law.⁸⁸ For the obligation of taking preventive operation measures to arise,

it must be established that the authorities knew or ought to have known at the time of the existence of a real and immediate risk to the life of an identified individual from the criminal acts of a third party and they failed to take measures within the scope of their powers which, judged reasonably, might have been expected to avoid the risk.⁸⁹

The Court has also added that the preventive operational measures

must be interpreted in a way which does not impose an impossible or disproportionate burden on the authorities. Accordingly, not every risk to life can entail for the authorities a Convention requirement to take operational measures to prevent that risk from materialising.⁹⁰

The innovation of Kurt v Austria is that the GC modified the Osman test regarding the nature of the risk required so that the obligation is triggered in the specific context of domestic violence. This specification will be addressed in the next section. Another innovation of Kurt v Austria is that the GC added a fourth positive obligation framed in this way:

the assessment of the nature and level of risk constitutes an integral part of the duty to take preventive operational measures where the presence of a risk so requires. Thus, an examination of the State’s compliance with this duty under Article 2 must comprise an analysis of both the adequacy of the assessment of risk conducted by the domestic authorities and, where a relevant risk triggered the duty to act was or ought to have been identified, the adequacy of the preventive measures taken.⁹¹

The Court has thus framed a procedural positive obligation that implies risk assessment.⁹² Given the high level of abstraction in which it is expressed, the relationship between this procedural positive obligation, and the substantive positive obligation of taking protective operational measures, is not entirely clear. As the above quotation suggests, the first is ‘an integral part’ of the second, which suggests that risk assessment is not a separate obligation. However, the question remains how the nature and level of risk can be assessed, if there is no obligation to assess risk in the first place. Below, I will show how this ambiguity of the nature of the procedural obligation is addressed with the concretisation of the obligation. I will also show its effects on the determination of breach.

Here, it is also relevant to observe that the obligation to conduct a risk assessment can be perceived as more specific than the obligation of ‘taking appropriate steps to safeguard’ life. Yet, I have placed both of them at Level 1. This inconsistency reflects an instability in the distinctions between the Levels. Yet, paragraph 159 of Kurt v Austria where the Court articulated the obligation of a risk assessment is in the part of the judgement entitled ‘General principles’. If it was limited to the domestic violence context, the risk assessment would have been formulated in the part of the judgement entitled ‘Positive obligations under Article 2 in the context of domestic violence’, which would justify its conceptualization at Level 2. In addition, the ambiguous relationship between the obligation of conducting the risk assessment and the obligation of taking protective operational measures also explains why I frame both of them at Level 1.⁹³ It is also pertinent to note that in its subsequent case law, the Court has invoked the obligation of the risk assessment in other contexts, not limited to domestic violence,⁹⁴ which additionally justifies the conceptualization of this obligation at Level one. An outstanding question is whether the risk assessment as a general obligation framed at Level one has been invoked by the Court in cases of non-deadly violence. The case law after Kurt v Austria in the area of domestic violence suggests an answer in affirmative.⁹⁵

As to non-deadly violence/harm outside the context of domestic violence, I found no judgements in this effect in the post Kurt v Austria case law.⁹⁶ It remains therefore to be seen whether the Court will expand the ECHR by formulating an obligation of risk assessment at Level 1 across all contexts and regardless of the deadliness of the harm. Such a development might be anticipated for two reasons. First, the obligation of conducting risk assessment is related to the obligation of taking protection operational measures (i.e. the Osman test). The latter has been certainly applied to forms of harm that are not deadly.⁹⁷ Second, the Court tends to expand positive obligations by drawing on standards developed in one context and applying them to a different context.⁹⁸ Despite these two reasons, caution is also warranted. The formulation of a general obligation of risk assessment at Level 1 regardless of the context and the deadliness of the harm might be an unreasonable expansion of positive obligations via encouragement of measures of risk aversion and pre-emptive actions aimed to avert or protect from risks. Societies might have to accept certain levels of risks in various contexts, which militates against the expansion of the positive obligation and, accordingly, against its articulation as a distinctive obligation at Level 1.

It is finally important to note that at Level 1 analysis, where very abstract and general positive obligations are framed, no determination of breach is made and the challenging questions about delimiting responsibility for omissions are not relevant. Rather, the objective is the formulation of some general principles regarding the objective of safeguarding life, which can be related to the constitutional role of the Court⁹⁹ to formulate some general standards for state conduct.¹⁰⁰

B. Level 2—General Positive Obligations under Article 2 in the Specific Context of Domestic Violence

After framing the above-mentioned four positive obligations at an abstract level, a second step can be identified in the Court’s reasoning. This step entails the concretisation of the obligations to the context of domestic violence.¹⁰¹ In particular, in Kurt v Austria, the GC affirmed that ‘special diligence is required from the authorities when dealing with cases of domestic violence’.¹⁰² Given the Court’s constitutional role, this affirmation seems to send a signal that more demanding positive obligations might be formulated. This will be better explained below with reference to the specification of the Osman test. The reference to ‘special diligence’ can also mean that each individual case of domestic violence that might raise an issue of state responsibility for omissions is examined in light of the wider societal problem of domestic violence. As the GC framed it, ‘[t]he issue of domestic violence – which can take various forms […] – transcends the circumstances of an individual case’.¹⁰³ This might imply that the reasoning is more contextual and abstract, not exclusively focused on the concrete factual circumstances and on the concrete causal links between specific omissions and harm. Such a contextualisation and abstraction of the reasoning might imply an easier finding of a breach when the case fits within some wider problems perceived as structural in society.

Having observed that cases of domestic violence are to be subjected to a more abstract review given the ‘special diligence’ required from the State, and that the issue of domestic violence ‘transcends the circumstances of an individual case’, we can now see how the Court concretizes the general positive obligations mentioned in Section 3.A in the specific context of domestic violence. The analysis in this section is limited to the concretisation of the obligation of taking preventive operational measures (the Osman test) and the obligation to conduct risk assessment, since these obligations were further specified by the GC in Kurt v Austria. In contrast, the obligation to develop an effective regulatory framework was not further specified. The reason is that the GC chose not to frame omissions related to the national legal framework as causative to harm, which made the further specification of this positive obligation unnecessary in the reasoning.¹⁰⁴ Section 3.C.(iv) will suggest possible reasons for this choice.

(i). The obligation to take preventive operational measures in the specific context of domestic violence

The Court specified the immediacy standard that the Osman test demands by noting that the obligation of taking preventive operational measures is triggered in ‘any situation of domestic violence in which harm is imminent or has already materialised and it likely to happen again’.¹⁰⁵ This means that the immediacy of the harm is not a requirement for the State to be under the positive obligation of taking protective operational measures. Rather, the likelihood for the harm to happen again suffices in the specific context of domestic violence. The Court clarified that the standard of immediacy has to be applied

in a more flexible manner than in traditional Osman-type situations, taking into account the common trajectory of escalation in domestic violence cases, even if the exact time and place of an attack could not be predicted in a given case.¹⁰⁶

The Court further specified the nature of the operational measures called for by providing three examples: risk management plans, coordinated support services for victims and treatment programmes for perpetrators.¹⁰⁷ As to pre-trial detention, the action invoked by the applicant, the GC held that this was not a measure that can as a matter of principle be used, unless it meets the demanding requirements of Article 5(1) that enshrine the right to liberty.¹⁰⁸

(ii). The obligation to conduct lethality risk assessment

As to the obligation to conduct risk assessment, the Court specified it as a ‘lethality risk assessment’. It noted that

in order to be in a position to know whether there is a real and immediate risk to the life of a victim of domestic violence, the authorities are under a duty to carry out a lethality risk assessment which is autonomous, proactive and comprehensive.¹⁰⁹

In contrast to the ambiguity as to the independent existence of this procedural positive obligation when framed by the Court at Level 1, the Level 2 analysis performed by the Court speaks in favour of an independent obligation. Three reasons support this conclusion. First, the judgement contains a distinct subsection entitled ‘Obligations relating to risk assessment’. Second, the first paragraph in this subsection refers to Article 51 of the Council of Europe Convention on Preventing and Combating Violence against Women and Domestic Violence. Article 51 is very clear to the effect that it establishes an obligation upon States to ‘ensure that an assessment of the lethality risk, the seriousness of the situation and the risk of repeated violence is carried out’. This is an independent procedural obligation, not linked to the result of the risk assessment or to any post factum assessment as to whether any measures were taken after the risk assessment. Third, the statement in the judgement that the ‘authorities are under a duty to carry out a lethality risk assessment which is autonomous, proactive and comprehensive’, also suggests an independent procedural obligation.

In addition to these three reasons, it is relevant to observe that the domestic violence case law after Kurt v Austria supports the position that the procedural obligation of risk assessment is independent.¹¹⁰ Yet, some deviations could still be identified,¹¹¹ which signals uncertainty. This might be an indication that in some cases the Court might still prefer to integrate the risk assessment exclusively within its Level 3 analysis as part of the Osman test. The consequence of such an approach will be that the concrete measure of risk assessment, even in the domestic violence context, will be narrowly related to the specific facts of the case and the importance of the omission to undertake such a measure tailored to the specific case. As a consequence, such an omission as a basis for the establishment of state responsibility is less likely to be assessed on its own, but as related to other omissions or other considerations (i.e. those reflected in the Osman test) as relevant in light of the specificity of the concrete case.¹¹² These reflections demonstrate the importance of identifying the level of concreteness at which measures that form the content of positive obligations are framed. A hesitation by the Court to formulate a measure at Level 2 might signal that the omission on its own to undertake this measure is less likely to lead to breach as assessed Level 3. In this sense, the formulations used at Level 2 affect the analysis at Level 3.

The uncertainty as to whether lethality risk assessment is a separate and independent obligation was of limited significance in the Level 2 analysis itself, since similarly to what was explained regarding Level 1, no determination of breach is made at Level 2. Although the reasoning is more concretely relevant to the facts of the specific case, the objective of the pronouncements made at Level 2 is the formulation of general principles for furthering the objective of safeguarding life, which objective can be also related to the Court’s constitutional role. The question of breach rather arises at Level 3 analysis, which reflects the most concrete level of framing positive obligations. Analytically, this is the most challenging level, where the analysis zooms in on the specific facts and where individual justice guides the reasoning.¹¹³

C. Level 3—Concrete Positive Obligations in Light of the Specific Case of Domestic Violence

This section will show how positive obligations are articulated in the reasoning, so that a determination about breach can be made in light of the concrete facts of the case. In relation to each positive obligation (taking preventive operational measures to protect the applicant, conducting lethality risk assessment, taking preventive operational measures to protect the children and adopting effective regulatory framework), it is highlighted that various concrete articulations are possible. Although such articulations are necessary in the reasoning, they also imply that the Court might adopt a rule-making role, which might undermine state discretion. Techniques can nonetheless be identified in the reasoning for obscuring this role so as to avoid the perception that state discretion might be undermined.

(i). Obligation to take preventive operational measures to protect the application

Given the history of domestic violence the applicant was subjected to, it was easy to conclude that the national authorities were under the obligation to take preventive operational measures to protect her. The key question was rather what concrete measures this obligation implied.

1. Merge of the two steps: Imposition of an obligation and determination of breach

As explained in Section 2, if the distinction between the existence of an obligation and its breach is to be maintained, such measures will have to be specified in advance in the Court’s reasoning and then an assessment made whether the measures were actually undertaken. This is difficult to perform for two reasons: practical and institutional. On the practical side, it is difficult to think about and enumerate all possible measures that could or should have been undertaken when a situation like the one of the applicants arises. This relates to the observation made in Section 2 that a State may perform multiple omissions, and such omissions have no definitive counterparts.

Even if it were practically feasible to enumerate all possible measures, the reasoning as a post factum review would also have to include an assessment of whether all of these measures could have contributed to preventing the harm, so that the requirement for causation is fulfilled. If they could not contribute, these measures should not be part of the positive obligation. Even if these thinkable measures could have contributed to prevention, it also needs to be considered whether it was reasonable to undertaken them. In sum, it would be a challenge to practically frame the reasoning in a way that implied specification of the measures in advance, so that the content of obligation was articulated in the beginning of the analysis. In addition, it might make little sense to specify measures, as part of the formulation of the content of the positive obligation, when it is not clear whether these measures could have prevented the harm and/or were reasonable to undertake.

Besides this practical challenge, there is also an institutional aspect related to the relationship between the Court and the State Parties. The starting point in this relationship is that States have discretion as to what measures to undertake to fulfil positive obligations. The Court has consistently confirmed this in its case law:

the choice of means for ensuring the positive obligations under Article 2 is in principle a matter that falls within the Contracting State’s margin of appreciation. There are a number of avenues for ensuring Convention rights, and even if the State has failed to apply one particular measure provided for by domestic law, it may still fulfill its positive duty by other means.¹¹⁴

The Court has also stated that it is not its role ‘to replace the national authorities and to choose instead of them from among the wide range of possible measures that could be taken to secure compliance with their positive obligations’.¹¹⁵ Therefore, States have at their disposal a wide range of alternatives, and it is difficult to specify the measures in advance. Such an initial specification might also bring the Court very close to taking on the role of rule-maker,¹¹⁶ which also explains the merging in the analysis of the two analytical steps, i.e. the imposition of an obligation and the determination as to its breach. Given the integration of these two steps, the judgement is tailored to the specific facts, which creates the impression that it does not entail imposition and definition of new rules, but rather the application of rules to facts.

In sum, since the measures cannot be specified, the obligation also cannot be specified. The whole assessment of state responsibility therefore is reduced to the question of breach: Did the State breach an obligation that has never been specifically articulated in advance and whose content and scope is not known, and in this sense cannot be specified in advanced either? In other words, the questions whether there is an obligation and what its content and scope might be collapsed into the question of breach.

2. Ways of specifying the obligation

How has the Court more specifically approached these conceptual difficulties? In particular, how has it, in its reasoning, dealt with the conceptual challenge that omissions have no definitive counterparts but rather introduce hypothetical and counterfactual analyses? Three interrelated ways can be identified in the Court’s reasoning.

Review of the measures actually undertaken

First, in Kurt v Austria, the Court enumerated all the measures actually undertaken by the national authorities when the events were unfolding and concluded that these measures ‘demonstrate that the authorities displayed the required special diligence in their immediate response to the applicant’s allegations of domestic violence’.¹¹⁷ This way of reasoning in the judgement shows that it is possible to know whether the obligation has been fulfilled only post factum in the context of a post factum review, like the one performed by the Court. It also shows that the standard of ‘required special diligence’ cannot be specified in advance before the post factum review. This standard has no initial and independent content. It cannot exist independently of the concrete facts, which further explains the collapse of the two analytical steps (i.e. the imposition of an obligation and determination of breach) in the reasoning.

Review of the measures explicitly invoked by an actor

A second way can be identified in the reasoning in Kurt v Austria for approaching the difficulties arising from the absence of an advance specification of the content of the obligation (i.e. the concrete measures) and the related absence of a conceptual distinction between the obligation and its breach. This second way implies framing the reasoning around the omissions specifically invoked by the applicant or other actors (e.g. third-party interveners). This means that the positive obligation is specified by the applicant’s proposal as to what measures should have been undertaken. In this way, the applicant makes the omission cognizable.¹¹⁸ For example, the applicant in Kurt v Austria complained about the choice of the measures taken by the national authorities: she argued that the perpetrator should have been taken into pre-trial detention. If the reasoning were to be framed around this argument, then the specific question the Court had to address would be expressed in this way: Did Austria have a positive obligation to take the perpetrator into pre-trial detention to protect the child’s right to life under Article 2 ECHR? A negative answer will imply simultaneously the absence of an obligation and the absence of a breach. A positive answer will necessarily imply an existence of an obligation and simultaneously its breach (since the husband was not taken in pre-trial detention).¹¹⁹ It follows then that the imposition of an obligation and the assessment of its breach converge again as analytical steps. This also explains why the Court avoids framing in its reasoning the positive obligation in such concrete terms in the very beginning of its Level 3 analysis but rather that the reasoning focuses on the question of breach.

Despite these two ways used by the Court to practically address the conceptual difficulties in framing the concrete positive obligations, it also needs to be acknowledged that multiple alternatives to those invoked by the applicant (or other actors) could be specified, so as to make an omission cognisable. For example, the GC judges who dissented invoked other alternative omissions.¹²⁰ The question that arises then is whether the examination of state responsibility should include these identifiable alternatives and, more generally, how to restrict the scope of such alternatives as can be included in the analysis.¹²¹ An institutional problem also arises: if the establishment of state responsibility includes consideration of multiple alternatives framed in very concrete terms, this can amount to indicating the precise measures that should have been undertaken by the State. This is difficult to square with the starting point, already mentioned above, that States have discretion as to what measures to undertake to comply with their positive obligations.

The use of qualifiers such as ‘reasonable’

Due to this tension, it is understandable why the Court refers to malleable terms like ‘sufficient’, ‘adequate’, ‘effective’ or ‘reasonable’ to characterise the measures that form the content of positive obligations.¹²² These qualifications serve simultaneously three interrelated objectives. First, they enable the Court to deal with the multiplicity of possible proposals as to which measures could have been taken. Second, in the absence of an initial standard (i.e. the concrete protective measures that form the content of the positive obligation), they analytically enable the reasoning of the Court. Third, the usage of these terms serves the objective of preserving state discretion as to what concrete protective measures should be undertaken. Below, I will explain these three objectives in more detail.

As to the first objective, flexible terms, such as ‘sufficient’, ‘adequate’ or ‘reasonable’ serve a particular function: they enable the judges to provide a reasoning when faced with abstract standards, such as those at Levels 1 and 2. In other words, these flexible notions enable the Court to reason and to reach a conclusion (i.e. breach or no breach of the ECHR) when faced with the choices about how to integrate facts from the social reality with the abstract standards, when multiple alternative paths are possible as to the course of conduct that could (or should) have been adopted by the State.

As to the second objective, the usage of qualifiers such as ‘reasonable’ to describe the measures that form the content of the positive obligation creates the impression that there are two analytical steps in the reasoning: first, the description of the measures (i.e. the content of the positive obligation), and second, an assessment whether the measures have been performed (i.e. the breach of the positive obligations in case of non-performance). Yet, since the meaning of what is reasonable cannot be pre-determined,¹²³ and since various considerations can be relevant for determining reasonableness,¹²⁴ the distinction between these two steps is difficult to maintain. Analytically, the two steps collapse into each other, since what is ‘reasonable’ can only be assessed at the second step.

As to the third objective, the reference to reasonableness, sufficiency and adequacy to characterise the measures legitimatizes the Court’s decision-making process and the discretionary power of the judges to define rules.¹²⁵ As Corten has explained, ‘the notion of reasonableness enables judges to present their decisions and motivations, often of their own creation, as perfectly in line with the intention of States’.¹²⁶ In this way, qualifiers such as ‘reasonable’ obscure the rule-making role of the Court. In sum, the flexible terms used in the framing of positive obligations mask the rule-making function of the Court that it inevitably adopts when it applies abstract standards to concrete facts, and hide multiple and divergent alternatives that can be used to achieve the objectives pursued by these abstract standards (e.g. safeguarding the right to life).¹²⁷

This does not mean that there is a complete and categorical conflation of the actual formulation of the standards (i.e. the formulation of the concrete positive obligation) and their application (i.e. the determination of breach of the concrete obligation), leading to an assessment of breach that is completely unpredictable. Given that the Court has framed positive obligations at Level 1, has adapted them to some contexts (what I frame as Level 2) and has consistently reiterated them in its judgements for a long time, it is hard to say that there is a complete conflation. In addition, the Level 3 analysis can indirectly provide examples of more concretely framed obligations that can be relevant for future cases. In this sense, every documentation of a breach at Level 3 is a way of signalling what more concrete measures might be considered as required for complying with positive obligations. All this demonstrates the value of having the different levels of concretization of the obligations and the importance of clarifying them, which is the objective of this article. It also shows the importance of better understanding, specifically the last level of concretization (i.e. Level 3) and, in particular, the choices that can be made within this concretization. These choices that inevitably have to be made in the reasoning challenge the distinction between an initial existence of an obligation with some initially defined scope and content, on the one hand, and application of the obligation, on the other. In this sense, despite the cogent arguments for separating the definition of positive obligation and their application,¹²⁸ rule making and rule application are analytically difficult to isolate as self-confined steps in the reasoning.

Going back to the facts in Kurt v Austria, a relevant question related to the identification of the concrete measures (i.e. the content of the positive obligation of taking protective operational measure) is whether it is possible to formulate these measures in concrete terms, when no risk assessment has been performed and, in this sense, the risks have not been specified enough. This brings us back to the positive obligation of conducting risk assessment and the question of its independent nature when the Court precedes to Level 3 analysis.

(ii). Obligation to conduct lethality risk assessment

Three steps will be undertaken to better understand the positive obligation of conducting risk assessment. The first focuses on the question whether breach of the obligation is dependent on the result of the risk assessment. Second, I will explain the implications for the role of the Court, if the obligation is understood as being one of means as opposed to one of result, where the criterion for the distinction between obligations of means versus obligations of result is the concreteness in their framing. Third, I will explain the implications, if the criterion for this distinction is the efforts performed.

1. A procedural obligation dependent on the result

The facts in Kurt v Austria showed that the national authorities conducte risk assessment in relation to the applicant. The GC had to review only whether this risk assessment met the qualitative standards of being autonomous, proactive and comprehensive. These standards were found to have been complied with,¹²⁹ a relatively easy conclusion to reach given the facts and the outcome (i.e. the applicant herself was not killed). The more difficult question for the GC concerned the obligation of conducting lethality risk assessment regarding the children. The Court ‘solved’ the question by observing in paragraph 205 that

While it is true that no separate risk assessment was explicitly carried out in relation to the children, the Court considers that on the basis of the information available at the relevant time this would not have changed the situation, […].¹³⁰

Two observations are due here. The first concerns the nature of the procedural positive obligation of conducting a lethality risk assessment as framed at Level 3 analysis. In particular, in stark contrast to the Level 2 analysis (see above at Section 3.B.(ii)), this procedural obligation is not framed as an independent one at Level 3 analysis. This implies that since there was no separate positive obligation to conduct risk assessment regarding threats to the life of the children, the question of breach did not need to be engaged with. If it had been, an easy finding of breach would follow.

The second observation regarding paragraph 205 concerns the Court’s reasoning that even if risk assessment in relation to the children had been carried out, ‘this would not have changed the situation’. In other words, even if the authorities had done a separate risk assessment, the result would have been the conclusion that the children were not at risk. The Court clarified that

The authorities could legitimately assume that the children were protected in the domestic sphere from potential non-lethal forms of violence and harassment by their father to the same extent as the applicant, through the barring and protection order. There were no indications of a risk to the children at their school let alone a lethality risk.¹³¹

Indeed, given the facts of the case, it seems correct that there were no indications that could make the national authorities aware that the children might be exposed to a lethal risk. However, this refers to the result of the risk assessment, not to the question whether the national authorities conducted a risk assessment in relation to the children. It was clear that they did not.¹³² If conducting a risk assessment is a separate obligation that arises in the context of domestic violence as suggested by the Court in its Level 2 analysis, failure to fulfil it should lead to a breach. However, when the Court concretely formulated the obligation at Level 3 to the specific facts to assess breach, it watered it down by making the result a relevant consideration. This illustrates how the reasoning is pulled in different directions. On the one hand, the Court aims with the judgement to establish some general standards of state conduct by formulating positive obligations at the abstract level (such as Levels 1 and 2). On the other hand, when these obligations are concretized and applied to the particular facts, they can be narrowed down. For this purpose, the overall reasoning includes ambiguities as to the nature and content of the obligation (i.e. is it one of means or of result), which ensures the scope of manoeuvre for the Court when determining whether the obligation has been breached. The post Kurt v Austria case law has not been helpful in resolving this ambiguity.¹³³

As suggested above, the reasoning in Kurt v Austria implies that the assessment, whether there was a positive obligation to conduct risk assessment and whether it was breached, is linked with the result (i.e. the conclusion that the children were not at risk). At face value, this seems difficult to square with the general principle established in the case law that positive obligations are obligations ‘of means and not of result’.¹³⁴ At this juncture, a brief foray into the meaning of this characterisation of positive obligations as obligations of means is appropriate. It is also relevant to reflect upon how this dichotomy (i.e. obligations of means versus obligations of result) can be applied to procedural obligations, and how it affects the way in which breach is determined.

The distinction between obligations of means versus obligations of result is suggestive of the taxonomy between obligations of conduct and obligations of result used in general international law.¹³⁵ Importantly, there are two possible meanings that can be attached to this taxonomy. These two meanings will be clarified below, and their explanatory relevance to the positive obligations under the ECHR will be elucidated.

2. The distinction between obligations of means and obligations of result with reference to the level of concreteness in their framing

The first meaning of the distinction between obligations of means and obligations of result refers to the concreteness in the framing of the obligations. The distinguishing criterion thus ‘rests on the extent to which international law encroaches upon the state machinery by instructing state organs to conform to a particular behaviour’.¹³⁶ In this sense, obligations of conduct ‘specify the means of the course of action to be taken by the state’.¹³⁷ Obligations of conduct/means are therefore more specifically framed, or more concrete. In contrast, obligations of result are more abstractly framed and respect the discretion of the State how to achieve a result.¹³⁸ Given the relatively concrete framing of the obligation to perform risk assessment, this obligation can be classified as one of conduct/means, since it requires the State to adopt a particular course of conduct. This also fits well with the nature of procedural obligations as obligations that require specific steps to be taken that enable a particular result (i.e. a particular decision).¹³⁹ The decision itself is not under review, but the steps and whether they comply with certain procedural guarantees.

The determination as to whether obligations of means/conduct are breached can be regarded as easier to make since such obligations require the State to undertake certain concrete measures. Risk assessment can be viewed as an example of such a measure and ‘the mere fact of not adopting that measure constitutes in itself a breach of the international obligations in question’.¹⁴⁰ If this reasoning is followed and, as already suggested above, the mere fact that Austria did not conduct risk assessment regarding the children should lead to an easy finding of a breach. In this sense, obligations of means/conduct can be assessed as more demanding since they imply that international law directs States as to what concrete measures to undertake in this way undermining their internal discretion.¹⁴¹

In contrast, the determination of breach of obligations of result is less easy. States are free to choose the measures (i.e. the concrete conduct) to achieve the result. As Ollino has explained, the breach of an obligation of result requires ‘a failure of the state, in concreto, to achieve the desired result. In other words, the mere adoption by the state of measures abstractly in conflict with the result set by the obligation was not sufficient for determining responsibility’.¹⁴²

The crucial question that arises here is the identification and the articulation of the result in the formulation of the obligation. An equally crucial question is the distinction between means and result. Applied to Kurt v Austria, these questions can be framed in the following way: Is the result in question the saving of the child’s life? Is the result in question putting the authorities in a position that could enable them to know about risks to life and risk assessment is the means that can help in the achievement of this result? Is the result in question the taking of protective operational measures? If the GC reasoning in Kurt v Austria is followed, all of these questions should be answered in the negative. Rather, the result that the reasoning refers to is the knowledge about risk. Pursuant to the GC, no knowledge about risk for the life of the child would have been gained, even if risk assessment had been performed. Thus, the reasoning arguably merges elements that characterise the obligations of means (i.e. risk assessment is a concrete measure and, in this sense, the content of an obligation of means) and obligations of result (i.e. the gaining of knowledge about a lethality risk, a result that could not have been achieved).

This conclusion, however, is contingent on how the Court chooses to formulate the result. If the result is formulated as ensuring that the authorities are in a position that they could have known better, a breach could have been found. This possibility for reformulation suggests that the obligation to conduct a risk assessment as applied at Level 3 in the reasoning in Kurt v Austria does not fit within the above-described classification of obligations (means versus result). The classification therefore might not be very useful in guiding the process of reasoning leading to a conclusion of breach or no breach. This finding is in line with Crawford’s observation that the distinction is ‘not a dichotomy but a spectrum’.¹⁴³ Tomuschat has also observed that obligations can combine both aspects—means and results—‘to such a degree that the different components can hardly be separated from one other’.¹⁴⁴

The distinction still has some analytical value, however. Using it as a reference point allows us to conclude that the obligation to conduct risk assessment appears to be an obligation of means, since it is concretely formulated. As such, an easy determination of breach could have been made. However, such a determination could imply direct intrusion into state discretion, since it entails a formulation of a concrete measure to be undertaken at national level. The intrusion was avoided by reference to a result. The result used as a point of reference in the reasoning was the gaining of relevant knowledge by the Austrian authorities. This result would not have been achieved by the state authorities, even if the concrete measure of risk assessment had been taken.

3. The distinction between obligations of means and obligations of result with reference to the efforts

A second meaning, not related to level of concreteness in the formulation of the obligation, can be attached to the dichotomy between obligations of means/conduct versus obligations of result. According to this second meaning, obligations of means/conduct can be regarded as less demanding since they require of the State only to endeavour towards the pursuit of a certain result without necessarily achieving the result.¹⁴⁵ In this sense, obligations of conduct are obligations of efforts, while obligations of result demand the achievement of a certain result, and in this sense are more demanding. It is in this sense that the Court uses the dichotomy between obligations of means and obligations of result. Positive obligations are therefore obligations of efforts.¹⁴⁶

However, procedural positive obligations, such as the obligation to conduct risk assessment, pose a conceptual challenge. If they are separate and independent obligations,¹⁴⁷ they can be viewed as obligations of result, the result being the risk assessment itself. If this result is not achieved, as in Kurt v Austria (no risk assessment was performed regarding the children), breach should follow. This is not how the Court framed its reasoning in Kurt v Austria, though.

The procedural obligation can, however, be linked to the obligation of taking protective operational measure, which entails that the former is not a separate obligation. It is rather part of the efforts, and breach will be assessed accordingly.¹⁴⁸ In other words, as noted above, obligations of efforts are less demanding and breach less likely. This seems to be the approach adopted by the GC in Kurt v Austria in its Level 3 analysis. The reference to a result (i.e. no relevant knowledge would have been gained anyway, even if the authorities had carried out a risk assessment) in the GC’s reasoning does not aim to modify the nature of the obligation (as an obligation of means and thus of efforts). The reference to this particular result rather reflects the post factum assessment by the Court of the efforts.

If, however, the result, the aim of the efforts, were to be framed differently, the reasoning would be different. Specifically, the aim of the efforts could be framed as placing the state authorities in a position to know better about risks. Risk assessment could still be conceptualised as part of these efforts.¹⁴⁹ If, however, the result/aim is framed in this way, the benefit of the hindsight would be less relevant. It would be less relevant that in retrospect it can be concluded that even if the authorities had placed themselves in a position to know better, they would not have gained relevant knowledge. This different framing of the result might imply that it is less likely that the Court will benefit from a post factum assessment.

All of the above demonstrates how obligations can be framed differently and how this framing affects the determination of breach. The analysis also shows that the framing of positive obligations does not necessarily fit into specific classifications. The taxonomy of obligations of means, as opposed to obligations of result, has analytical value,¹⁵⁰ since it is useful to better understand positive obligations as developed by the Court. Yet, it is crucial to also deconstruct the taxonomy and to understand what stands behind it. Once this is done, it becomes clear that there are different aspects—the level of concreteness in how the obligation is framed, the related intrusion into State discretion, the distinction between means (i.e. efforts) and objectives (i.e. results), the choice made as to how the result is framed and the role played by the benefit of the hindsight. The assessment of breach includes choices as to which of these different aspects should be given prominence. The choices can affect the conclusion (i.e. breach or no breach).¹⁵¹

(iii). Obligation to take preventive operational measures to protect the children

The choice as to which aspect to highlight and how to concretise was also essential when the GC considered the obligation of taking preventive operational measures to protect the children. As mentioned in Section 3.B.(i), this obligation arises upon real and immediate risk. The concretisation of the immediacy standard was already explained in the same section. Other aspects of the obligation, however, can be also objects of concretisation. In particular, the risk could be circumscribed to specific locations (geographical delimitation of the risk) and/or to specific timing (temporal delimitation of the risk) and/or as being of specific kind/severity/category (category delimitation). For example, in Kurt v Austria, the GC noted that

[…] the Court agrees with the Government that, on the basis of what was known to the authorities at the material time, there were no indications of a real and immediate risk of further violence against the applicant’s son outside the area for which a barring order had been issued, let alone a lethality risk.¹⁵²

These words suggest a category delimitation, since the risk was not framed more generally as risk of harm but as lethality risk. A geographical delimitation of the risk was also used in the formulation of the obligation, since it was noted that ‘[t]here were no indications of a risk to the children at their school’.¹⁵³ The reference to certain areas and to ‘private or public spaces’ also suggests a geographical framing of the risk.¹⁵⁴

The use of such concretisations in the formulation of the obligation is analytically necessary so that the Court applies the more abstract standards (i.e. the formulations at Levels 1 and 2) to concrete facts. The crucial point here is that when the Court reasons, it can choose which concretisations to use when it articulates the obligation and, accordingly, which aspects to highlight. In Kurt v Austria, the obligation was framed narrowly with reference to the above-mentioned delimitations, which made it easier to conclude that there was no breach.

A clarification is immediately due here. Given the above-mentioned, geographical, temporal and category concretisations, the GC concluded that ‘there was no obligation incumbent on the authorities to take further preventive operational measures specifically with regard to the applicant’s children, whether in private or public spaces, such as issuing of a barring order for the children’s school’.¹⁵⁵ This quotation suggests that there was no obligation (as articulated in the specific way chosen by the Court) in the first place, and the question of breach is therefore irrelevant. At the same time, the reference to ‘further preventive operation measures’ suggests that there was actually an obligation to take preventive operational measures. The measures actually taken were, however, assessed as adequate by the Court.¹⁵⁶ All of this again shows that the choice of how the obligation is framed and articulated is crucial, first, for the distinction between the existence of an obligation and its breach, and second, for the establishment of breach.

The more the obligation is framed with reference to geographical, temporal and category concretisations, the more the finding of a breach will entail instructions to the national authorities as to the measures that should have been taken. This was not an issue that arose in Kurt v Austria, since no breach was found. However, it can be observed in other cases, where the conclusion was different and where the framing of the obligation contained various concretisations.¹⁵⁷ This can be seen as an intrusion in state discretion and erosive of the principle that States have discretion to choose the concrete measures to fulfil their positive obligations.¹⁵⁸

(iv). Obligation to adopt effective regulatory framework regarding barring orders

The obligation of taking preventive operational measures (the Osman test) is invoked when there is a concrete victim exposed to a specific kind of risk (defined as ‘real and immediate’) and omissions can be identified that, if avoided, could have prevented this specific kind of risk.¹⁵⁹ Omissions can also be identified, however, that are not of an operational, but of more systemic and structural nature pertaining to the national regulatory framework.¹⁶⁰ The risks that they pose might not be ‘immediate’ and the victims might not be concretely identifiable. Instead, the applicant in a specific case might happen to be a representative victim of some systematic and structural omissions. The analytical question that arises in this context is how to formulate these omissions, and relatedly, what level of specificity to use in formulating positive obligations.

As mentioned in Section 3.A, the GC in Kurt v Austria did not articulate an omission with reference to the national regulatory framework. The applicant, however, did formulate such omissions. First, the Security Police Act, as worded at the time of the events, was arguably deficient since it did not allow the police to extend the barring order to premises outside the vicinity of the home. Second, there was no requirement under Austrian law at the time of the events that information regarding domestic violence be shared with the children’s school.¹⁶¹ In the GC’s judgement, these omissions were mentioned as part of the assessment whether Austria had to take additional protective operational measures regarding the children. As clarified in Section 3.C.(iii), the GC reasoned that the respondent State was not under such an obligation.¹⁶²

The examination of these omissions through the prism of the Osman test implies that the omissions were not reviewed in abstract but rather in relation to their concrete effects on concrete victims. This meant that the review of the Court included only an assessment whether the legislative omissions were contributory to harm to the concrete victims. This way of applying the Osman test is understandable, given the nature of the obligation of taking protective operational measures. Specifically, the obligation is framed in relatively specific terms, with concrete identification of victims who arguably need to be protected and perpetrators who arguably need to be restrained. This level of specification also implies that the finding of a breach depends on more clearly on discernible causal links between the harm sustained by the concrete victims and the measures (i.e. the content of the obligation) meant to protect by restraining the perpetrators.

When omissions are reviewed through the prism of the positive obligation to adopt an effective regulatory framework, however, there is a wider scope for flexibility in the articulation of the causal links and, consequently, in the formulation of the positive obligation. The starting point for explaining this flexibility is the following principle established in the case law: ‘the Court reiterates that its review of the domestic regulatory framework is not an abstract one, but rather one that assesses the manner in which it affected the applicant in the specific case’.¹⁶³ This means that the Court has framed its task as examining the concrete application of the regulatory framework, rather than some general compliance. The Chamber in Kurt v Austria started its analysis as to whether the State had failed to adopt effective regulatory framework by reiterating this starting point. It is helpful to reflect here what it means that the review is not an abstract but a concrete one. The concrete review meant that the Chamber’s reasoning zoomed in on the implications of the legislation for the specific person and the specific harm sustained by this person. The Chamber in Kurt v Austria thus reasoned that the deficiencies in the legal framework did not sufficiently contribute to the specific harm (i.e. the killing of the child). The Chamber motivated the weak causality with reference to two main considerations. First, the authorities did not know about the risk to the child’s life, a motivation that reveals the intertwinement between the factors of causation and knowledge. Second, the applicant had at her disposal alternative measures enshrined in the legal framework: getting a temporary restraining order from the competent district court under the Enforcement Act, which could have banned her former husband from public places beyond residential premises.¹⁶⁴

This concrete review implied a specific framing of the positive obligation and its content. The questions the Chamber responded to were the following: Was Austria under the obligation to issue a barring order for the school to save the life of the child? Was Austria under the positive obligation to share information about the domestic violence with the school to save the life of the child?¹⁶⁵ This concrete framing of the obligation implies that the decisions by the national authorities in relation to specific individuals (in this case, the children) are reviewed. These decisions are indeed based on implementation of the general national regulatory framework. However, this framework as a whole, and any abstract defects that it might have, were not reviewed in the reasoning.

An alternative framing of the positive obligations is also possible, however. The obligations can be framed in a way that would demand a review of the reasonableness of the national regulatory framework as a whole.¹⁶⁶ This implies a more abstract framing of the obligation, where causation and knowledge recede into the background and reasonableness is placed in the foreground. If we take the facts of Kurt v Austria as an example, a more abstract framing would imply that the following question is placed at the heart of the analysis: Is it reasonable to have a national legislation that does not allow the extension of barring orders to premises outside the vicinity of the home and does not allow sharing of information with schools regarding domestic violence? Such an abstract framing implies that whether to what extent and how the omissions in the national legislative framework affected concrete applicants and contributed to their specific harm (or risk of harm) becomes less important in the reasoning.¹⁶⁷ Knowledge about risk of harm to the concrete applicants also becomes less important.

An abstract framing of the positive obligation thus implies more flexibility in the reasoning since it leads to an assessment of the overall reasonableness of the national legal framework. The causal effects between any rules or absence of rules on the one hand, and harm sustained by the concrete applicant on the other, recede to the background. The abstract review also accordingly implies that the positive obligations are framed in a way that is more detached from the concrete case and the concrete harm.

To conclude, despite the affirmation that the State has ‘special diligence’ to prevent domestic violence (see Section 3.B), neither the Chamber nor the GC went so far as to formulate more abstract positive obligations that would imply a review of the national legislation in more abstract terms and thus a review of the overall reasonableness of the national legislation. It is worthwhile, however, to reflect upon the implications if such a formulation and review were attempted and if such breach is found. The implication would be a pronouncement by the Court that States have positive obligations under the ECHR to issue barring orders and to share information. The role of the Court would then come closer to that of a legislator. This would be controversial and explains why the Court might not be willing to take this path.

4. CONCLUSION

The State as an abstract organisational entity that has an instrumental value and that is assumed to be in control of its territory, commits multiple omissions that can be framed as contributory to harm. Which of them can lead to state responsibility is contingent on their identifiability, on how omissions are framed and relatedly on how any positive obligations upon the State to act are formulated.¹⁶⁸ As to their identifiability, what characterises omissions is that they do not have definitive counterparts. Positive obligations can therefore be complied with via multiple independently sufficient measures that States can choose from. This discretion implies multiple ways for formulating any positive obligation. These formulations can oscillate between being more general or more concrete. The general and the concrete formulations, and the oscillation between the two, serve different purposes. As demonstrated with reference to Kurt v Austria, the Court has framed a general positive obligation upon the State to protect the right to life (Level 1 of concreteness) and has specified it in the context of domestic violence by noting that ‘special diligence’ is required in this context (Level 2 of concreteness). The purpose of this general formulation is the establishment of some general guidance for state conduct. The issue of breach does not arise at these two levels.

This issue arises when, in its reasoning, the Court has to integrate facts from the concrete case, which is a challenge since multiple proposals are possible as to the concrete measures that could have been adopted by the State. At this level of analysis, the Court is faced with choices as to how to formulate concrete obligations (Level three of concreteness). The choice made influences the finding of breach. In addition, the more the choice tilts towards a concrete formulation of the obligation, the more the Court appears to assume the role of a rule-maker, which is in tension with States’ discretion as to what concrete measures to take to fulfil their positive obligations. Yet, more concrete formulations of the positive obligations are analytically necessary in the legal reasoning so that a conclusion about breach can be reached given the concrete facts in the case.

Four major conclusions regarding the choice of framing positive obligations at Level three emerge. First, the usage of flexible notions such as reasonableness enables the Court to perform its review task when faced with choices on how to integrate facts from the social reality with abstract rules and when multiple proposals are possible as to the concrete measures that could have been adopted by the State. Second, besides the inclusion of flexible notions such as reasonableness, other techniques for making choices in the framing of the obligations could also be identified. Such a technique is the mixture of the means and the result in the analysis, which can be observed in the different ways in which the Court chose to frame the positive obligation of conducting a risk assessment in Knut v Austria. Choices can be also made as to which results (e.g. the risk assessment in itself, better knowledge about risks or the conclusion that the children were not at risk) should be used as points of reference. Third, choices are also available as to whether any geographical, temporal or category concretisations will be used in the framing of the obligations. Finally, different choices are also possible as to the level of causation demanded between the harm sustained by the specific applicant and any alleged omissions. The lower this level, the more abstract the way in which the obligation can be framed.

Overall, this article has demonstrated the value in better understanding the different levels of concretisation of positive obligations and how these levels affect the determination of breach. The key role of Level three concretisation was highlighted since it is at this level where multiple choices of formulations and specifications of measures arguably in fulfilment of positive obligations are possible. At Level 3, these choices have to be circumscribed and the limitations chosen affect the outcome (i.e. breach or no breach of the obligation). The necessary availability of choices at Level 3 reveals that the isolation between rule making and rule application (i.e. the isolation between existence of an obligation and its breach) as separate steps is difficult to achieve. This difficulty is aggravated in the context of positive obligations in comparison with negative ones due to the wider compliance choices available in the context of the former.

AKNOWLEDGEMENTS

This article has been completed as part of my Wallenberg Academy Fellowship awarded by the Knut and Alice Wallenberg Foundation and the Royal Swedish Academy of Sciences. I would like to thank Johan Vorland Wibye, Ulf Linderfalk, Evan Criddle, Andreas Føllesdal, Matthew Williams Saul and Serde Atalay for their comments on earlier drafts. The anonymous reviewers’ comments have also helped me to substantively increase the depth of my analysis, for which I am very grateful. The usual disclaimer applies.

Footnotes

Author notes