Causation between State Omission and Harm within the Framework of Positive Obligations under the European Convention on Human Rights

Causation between State Omission and Harm within the Framework of Positive Obligations under the... Abstract The issue of causation has been surprisingly overlooked in the area of international human rights law. The objective of this article is to fill this gap by investigating how the ECtHR finds causal connections between harm and state omissions within the framework of positive obligations. By engaging with causation, this article seeks to partially address the widely voiced concerns about the indeterminacy that clouds positive obligations in the case law. Four main arguments are articulated. First, assessments whether the state knew, or ought to have known, about the (risk of) harm, whether demanding state action is reasonable and whether harm is caused by state failures, are merged and affect each other in the enquiry as to whether the state has failed to fulfill its positive obligations. Secondly, the level of state control structures lines of causation. Thirdly, since the question as to how much control the state should have could imply normative judgments in which the Court might not want to see itself implicated, and since empirical and epistemological uncertainly might hamper assessments of causation, the Court has recourse to techniques to avoid direct resolution of these normative issues and uncertainties. Two such techniques are discussed: domestic legality and national procedural guarantees. Finally, even in cases where omissions might be causative to harm, additional considerations might militate against finding the state responsible under the ECHR: reasonableness, no immediacy of the harm and no systemic failures. 1. INTRODUCTION Positive obligations have penetrated all parts of the European Convention on Human Rights (ECHR or the Convention) and there are no a priori limits to the contexts in which they may be found to arise.1 Due to their considerable importance, scholarly contributions have focused on different areas covered by positive obligations, such as domestic violence and other forms of interpersonal violence,2 environmental pollution3 or socio-economic assistance.4 Efforts to systematize positive obligations as developed in the case law have also been made.5 There has also been extensive discussion of certain principles that affect these obligations, such as subsidiarity,6 margin of appreciation7 and proportionality.8 The role of causation, however, has been hitherto largely neglected. No analysis has been offered as to how the Court (ECtHR) finds causal connections between harm and state omissions. Filling this gap is the aim of this article. The importance of addressing this blind spot relates to the general uncertainty that clouds positive obligations. The ECtHR has not proposed a general analytical framework for reviewing them and has explicitly refused ‘to develop a general theory of the positive obligations which may flow from the Convention.’ 9 Perhaps as a consequence, it has been observed that the ECtHR’s approach to positive obligations is incoherent and even arbitrary, which is not conductive to certainty and predictability.10 One is left with the impression that the Court simply makes in casu judgments when dealing with positive obligations, and that it is hard to direct and structure these by extracting distinctions as to the steps taken and the principles applied. The quality of the Court’s reasoning has also been criticized. It has been noted that the problem with positive obligations is that ‘their proper scope appears open-ended’ and the Strasbourg court ‘does not set general conceptual limitations’ for its interventions in developing them.11 It has been added that such an approach ‘put[s] the concept of positive obligations into disrepute.’12 By engaging with causation, this article seeks to partially address the above concerns. While the problem of indeterminacy cannot be eliminated13 and in fact, should not be conclusively resolved,14 there are certain elements identifiable in the judgments which determine the intensity of the positive obligations.15 In this contribution, I focus on the element of causation as also related to the elements of knowledge and reasonableness. Causation implies some nexus/proximity between the harm sustained by the applicant (harm that falls within the definitional scope of one of the protected rights) and the alleged omission by the state to ensure the right. The issue of causation has been extensively addressed in other areas of law, but surprisingly neglected in the area of international human rights law.16 However, causation is essential for understanding positive obligations and responding to the above mentioned concerns as to the elusiveness of their scope. Any engagement with causation has to start with the awareness that ascribing causality in human society is fraught with complexities.17 There has been a clear acknowledgment that causality by omission is hard to ascertain. National law has struggled with issues of causality by omission18 and philosophy has also struggled with this issue.19 A further problem is that causality by omission implies a counter-factual and speculative analysis. It might be possible to identify different omissions that might have causal connections to the harm.20 The role of normativity and policy considerations when determining causality has been also noted.21 The issue of causality is also fraught with difficulties from an evidential point of view since determining causality might be a highly factual process. The ECtHR is certainly confronted with all these challenges that can be also related to the uncertainties about the scope and the intensity of the positive obligations generated by the ECHR. In this article, I will show how the ECtHR approaches these challenges. Section 2 will elaborate on the question as to why causation is significant in the context of positive obligations under the ECHR. Since the establishment of causation is related to the determination whether the state knew or should have known about harm or risk of harm, Section 3 will discuss the interrelationship between causation on the one hand, and knowledge and foreseeability on the other. Even if harm or risk of harm is foreseeable, the Court can still verify the place of state omissions in the chain of events. For this purpose, however, no clear test of causation has been articulated, as Section 4 will show. Against this backdrop, I suggest that inspiration from other areas of the law on state responsibility could be useful. Section 5 thus draws a parallel with the rules on attribution in international law. Since these rules express lines of proximity, it is useful to assess their underlying justifications so that we better understand the linkages between harm and state conduct. More specifically, the rules on attribution are founded on the principle that control implies responsibility and the same principle can be extended in the context of positive obligations. Accordingly, the degree of control exercised by the state is essential for assessing lines of causation. This is also reflected in the case law of the Court. Since the question as to how much control the state should have could imply normative judgments in which the Court might not want to see itself implicated, and since empirical and epistemological uncertainly might hamper assessments of causations, the Court can resort to techniques that avoid the direct resolution of these normative issues and uncertainties. Section 6 will identify two such techniques: domestic legality and procedural guarantees. Finally, Section 7 will demonstrate that an additional corrective can be used, namely the test of reasonableness, which permeates the whole case law on positive obligations. The last implies that even in cases of clear causal connections between harm and state omissions, it might be unreasonable to expand the scope of the positive obligations to such as an extent as to lead to state responsibility. Section 7 will also discuss two other techniques for limiting state responsibility: the ‘real and immediate risk’ test and the requirement for systemic as opposed to incidental failures. A brief note regarding methodology is due. In the selection of case law, priority is given to judgments under Articles 2, 3 and 8 of the ECHR delivered by the Grand Chamber. I have also analysed Chamber judgments, including those that are given prominence in the existing literature as heralding important developments concerning positive obligations. Although there are areas that have remained uncharted in this article,22 the selected case law can adequately serve the purpose of exploring the role of causation in the context of positive obligations. The method employed here is to both reflect upon and explain the approach taken by the Court. At certain points, however, I do take a critical approach to the case law. I extend the limits of my descriptive approach in the face of some inconsistencies within the case law and inadequacies as to how the Court has grappled with some issues that demand more serious consideration in the future. 2. WHY DOES CAUSATION MATTER? Identifying the causation between harm and state omission is crucial for finding the respondent state responsible for that omission under the ECHR. This was clearly exemplified in L.C.B. v United Kingdom, a case demonstrative of a failure to establish this connection. The applicant sought to attribute her leukemia to her father’s exposure to radiation from atmospheric tests of nuclear weapons during his military service. She claimed that the failure by the state to warn her parents of the possible risk to her health caused by her father’s participation in the nuclear tests, and its earlier failure to monitor her father’s radiation dose levels, gave rise to violation of Article 2. The Court did not dispute that the respondent state was generally under a positive obligation to protect the right to life. Accordingly, it defined its task as determining whether ‘given the circumstances of the case, the state did all that could have been required of it to prevent the applicant’s life from being avoidably put at risk.’ The Court could not, however, establish a connection between any omission by the state and the disease from which the particular applicant suffered: ‘[I]t is clearly uncertain whether monitoring of the applicant’s health in utero and from birth would have led to earlier diagnosis and medical intervention such as to diminish the severity of her disease.’23Botta v Italy is another case where the absence of proximity was decisive. The applicant complained that the state had failed to take measures to remedy omissions imputable to private bathing establishments, which prevented disabled people from gaining access to the beach and the sea. The Court found that there was no ‘direct and immediate link’ between the measures sought by the applicant and his rights under Article 8 of the ECHR.24 Causality and its function in the case law of the Court have not received adequate consideration in the existing literature. Benedetto Conforti, a former judge of the Court, is among the few who have noted its importance: A conclusion that can be drawn from the Strasbourg case-law is that no violation is found in cases where there is lack of a causal link between the behaviour of the State and the event. The same reasoning applies to cases where the event is not ‘the immediate consequences of state behaviour’, i.e. where the State omission is too remote in the chain of the various circumstances which led to the final wrongful event. … Couched in simple terms, in the opinion of the present author, the causation, as it appears in the case-law of the Court concerning the obligations of prevention, is strictly linked to the ‘foreseeability’ of the wrongful event. It is an ex post test of ‘foreseeability’ of the event: even if the event was predictable there is still room, even after the wrongful event occurred, for verifying its place in the chain of events.25 Conforti correctly draws attention to the interdependence between foreseeability of the harm and the imputability of the harm to an omission by the state. States are certainly not omniscient entities and it cannot be expected from them to know about and foresee any potential harm that individuals might suffer. It must therefore be established that the state knew or ought to have known about the harm.26 Accordingly, the establishment of causation is related to the establishment of knowledge.27 3. KNOWLEDGE AND CAUSATION It is important to clarity two points in relation to the knowledge requirement. First, it is not necessary to prove actual knowledge; it suffices to demonstrate that the state should have known about the harm.28 Secondly, the knowledge requirement operates differently in different contexts. In some circumstances, knowledge that a specific individual might be harmed has to be established.29 In other circumstances, however, this will be too demanding, since the applicant might simply be a representative victim of some general failure in the national regulatory system.30 In these circumstances, it suffices to establish that the state knew or ought to have known that this system caused harm. It is not necessary to show that the state knew or ought to have known that the particular applicant would be negatively affected.31 It might be enough to demonstrate that the state knew or ought to have known about the source of the harm.32 How is knowledge of an organisational entity such as the state established? The mere existence of legislation against certain harms suffices for concluding that the state knew about these harms.33 The Court has also referred to the standard of ‘objective scientific research’,34 to various reports that might have been drafted at national level, and to various channels of communication between state institutions.35 The state is to a certain extent involved in many activities, such as issuing permits,36 which might also facilitate the establishment of knowledge. As Lavrysen has argued, knowledge is not a static issue, and the state should constantly anticipate new harms.37 On the other hand, the existence of knowledge and its accuracy might be contingent on state resources, which warns against unreasonable expectations concerning the state’s capacity to augur potential harms. This unavoidably has an impact in assessing whether these harms bear nexus with the state. In this context, the Court has not yet fully appreciated the following distinction: whether the question as to when the authorities ‘ought to have known’ of the existence of a risk can be answered by asking whether (i) they ought to have correctly assessed the risk based on the information they would have had if they had carried out their duties with due diligence or, instead, (ii) they ought to have known of the risk based on the information that was in fact available to them. The first alternative presupposes that the authorities actually had duties, which might be a premature conclusion since it might be also contingent on the reasonableness of imposing such duties.38 This can explain the reluctance of the Court to make fine distinctions and instead to bundle issues of knowledge, reasonableness and causation in its assessment of state responsibility in the framework of positive obligations. The merging of these elements will resurface as an issue at various other points below. Knowledge and causation are intertwined elements also because in some circumstances the degree of foreseeability of the harm might be such that it is from that alone that the requisite proximity between the harm and the state can be deduced. Still, it is helpful to maintain the distinction between the different elements, since in other circumstances it would make little sense to enquire what measures could have prevented the harm, if the state authorities did not know about the risk of such harm in the first place.39 4. THE STANDARD OF CAUSATION What standard of causation has been applied in the case law? Ugrekhelidze, a former judge of the Court, has noted: It must be ascertained whether the violation would have been avoided had the relevant positive obligations been duly honoured, i.e. whether that is a causal connection between the failure to honour them and the violation. This is necessary in order for the attribution of international legal responsibility to a State to have a solid objective legal basis.40 Framing causation as requiring an affirmative answer to the question whether ‘violation would have been avoided’, as proposed in the above quotation, is however problematic for the following reason. It appears that Ugrekhelidze assumes that the cause of the harm is the state omission and the consequence of this omission is the harm. These strict causality links are an erroneous reflection of the requirement for proximity between the harm and the state conduct as developed in the case law. The omission by the state might be just one factor contributing to the occurrence of the harm. The ‘but for’ test, which means that but for the state failure the harm would not have happened, has been explicitly rejected by the ECtHR.41 From the perspective of the individual who claims to be a victim of a human rights violation, it would be too demanding and, ultimately, it might be impossible to prove that if the state had adopted effective protective measures, the abuse would not have happened.42 In short, there is no requirement that but for the omission, harm would not have materialized. Yet as highlighted in Section 2 above, there needs to be a proximity between the harm and the state omission, and the Court has formulated the following standard to this effect: ‘A failure to take reasonably available measures which could have had a real prospect of altering the outcome or mitigating the harm is sufficient to engage the responsibility of the State.’43 Avoidance of the harm and its mitigation are formulated as alternatives, which leads to further relaxation of the standard. In addition, the undertaking of protective measures by the state might only have had a real prospect of avoiding or mitigating the risk of harm, which adds further flexibility to the causation analysis. For example, in O’Keeffe v Ireland, the enquiry was framed as to whether ‘effective regulatory framework of protection in place before 1973 might “judged reasonably, have been expected to avoid, or at least, minimise the risk or the damage suffered” by the present applicant.’44 The Court does not consistently refer to the ‘real prospect’ test in all of its judgments, though. Note in this respect that the Court uses different expressions in order to refer to the causation between the harm and any omissions. For example, in Vilnes and Others v Norway it framed the question as to whether harm is ‘caused’, ‘attributable’ or ‘imputable to any specific shortcomings for which he [the applicant] criticized the State.’45 This question was asked without considering the degree of likelihood that the absence of these shortcomings would have mitigated the harm or the risk of harm. In Budayeva and Others v Russia, the review was framed as whether ‘there was a causal link’ between the serious administrative flaws that impeded the implementation of the land-planning and emergency relief policies and the death and the injuries sustained by the applicants.46 The Court has also used the expressions ‘direct causal link’,47 ‘direct and immediate link’48 and ‘strong enough link’.49 The term ‘nexus’ has been also utilized: ‘[t]he combination of these factors shows a sufficient nexus between the pollutant emissions and the State to raise an issue of the State’s positive obligation under Article 8 of the Convention.’50 The expressions ‘due to’51 and ‘can be linked directly’ have also been used.52 In E. and Others v United Kingdom, the Court referred to the standard of ‘significant influence’: the failings of the relevant authorities disclosed in the case ‘must be regarded as having had a significant influence on the course of events.’53 In contrast, in Makaratzis v Greece, it was observed that the deficiencies in the legal and administrative framework had a ‘bearing … on the way in which the potentially lethal police operation culminating in the applicant’s arrest was conducted.’54 In Hiller v Austria, the Court found that ‘M.K.’s escape [from psychiatric hospital] and subsequent suicide had not been foreseeable for the hospital and was not therefore attributable to it;’ this is a formulation that echoes the above discussion about the interdependence between foreseeability and causation.55 It is doubtful whether these various terms reflect any differences in the substance of the analysis, varying scrutiny as to the proximity element and different standards of causation. The Court has not developed anything close to a consistent terminology. Overall then and despite the usage of the ‘real prospect’ standard that appears to offer a more principled approach, uncertainty pervades the case law. 5. CONTROL AND CAUSATION A. The Rules on Attribution Against the backdrop of the above tangle and intricacy in the case law, the question emerges whether it is possible to find some structure by drawing inspiration from other areas of the law on state responsibility. Within the framework of state responsibility, issues of proximity arise in contexts other than positive obligations.56 More specifically, the rules of attribution in general international law themselves reflect lines of proximity. Attribution as defined in the International Law Commission (ILC) Draft Articles reflects rules for connecting conduct to the state.57 The rationale behind these rules is ‘limiting responsibility to conduct which engages the State as an organization.’58 As I will show below, the role of proximity in the realm of positive obligations is very similar: limiting the responsibility of the state to circumstances where the state is engaged in the harm as an organization. It needs to be acknowledged, however, that once attribution is established, the causation between state action and harm is evident. In contrast, in the context of positive obligations and claimed omissions, the lines of causation might not be that easily discernible and might raise challenges. The rules on attribution connect agents and entities to the state. Conduct is attributable to the state when committed by its actual organs59 and de facto organs60 or by entities directed and controlled by the state.61 As to the first group of organs, the fundamental principle is that the state is responsible for the actions of its organs, even if they act ultra vires62 and even if they are no longer under the control of the state.63 That the state controls its organs is a normative assumption; the capacity to control and the actual control are irrelevant. The conduct of state organs thus might give rise to per se responsibility on the part of the state.64 As the ILC has framed it: ‘[t]he attribution of conduct to the State as a subject of international law is based on criteria determined by international law and not on the mere recognition of a link of factual causality [emphasis added]’.65 With regard to non-state entities, their actions may be attributable to a state if they act under the instruction, direction or control of that state. Attribution arises in this context because ‘there exists a specific factual relationship between the person or entity engaging in the conduct and the State’ and ‘a real link between the person or group performing the act and the State machinery’.66 The ILC Draft Articles do not specify the level of control required, which has led to controversies. The International Court of Justice has applied the test of ‘effective control’; while other adjudicative bodies have endorsed the standard of ‘overall control.’67 This debate need not to detain us here. It might come as a surprise that I mention the rules on attribution in the context of positive obligations. While these rules are certainly of importance in the context of negative obligations, they are generally perceived as irrelevant when the case is formulated as one involving a failure to fulfill positive obligations.68 The obligation to ensure human rights does not require a determination that the actual harm is attributable to the state in the sense of the ILC Draft Articles on State Responsibility69 and the previously mentioned rules of attribution. The triggering of positive obligations and the scope of these obligations are contingent on the primary obligations at stake, which are not a subject of the law on state responsibility as such. However, it is still relevant to engage with attribution since the rules of attribution under international law articulate lines of proximity. They express relationships of directness and immediacy between the act of the state and the harm. It is meaningful to consider the justifications and the theoretical underpinnings of these relationships, so that we better understand the linkages between harm and state conduct in the form of omission. B. The Role of Control and the Extension of the Logic of the Rules on Attribution The rules on attribution seek to establish a nexus between the state and the agent who caused the harm. The status of the agent is thus of importance and the harm caused by him/her is directly attributable to the state. Even if the conduct is in breach of the national legislation or the state agent exceeds the authority granted by national law (ultra vires acts),70 attribution is still established as long as the organ acts within its capacity.71 This must be differentiated from cases ‘where the conduct is so removed from the scope of their official functions that it should be assimilated to that of private individuals, not attributable to the state.’72 In contrast, Article 8 of the ILC Draft Articles highlights the question of ‘direction and control.’ It clarifies that ‘such control will be attributable to the State only if it [the State] directed or controlled the specific operation and the conduct complaint of was an integral part of the operation.’ However, as observed by Evans, this is not the way the ECtHR has extended the scope of the obligations under the ECHR. It is rather positive obligations that have served this purpose.73 For example, there might be situations where it is not really possible to show that the state has directly caused harm through its agents or entities under its control.74 Two basic principles follow from the above. First, acts of state organs are attributable to the state and the issue of control seems to be immaterial.75 Secondly, when the first principle is not applicable since, for example, none of the requirements in Articles 4, 5 and 6 of the ILC Draft Articles is fulfilled, the second principle is triggered: control implies responsibility.76 This principle is not sidelined once a case is framed as involving positive obligations. On the contrary, control by the state (not in the sense of Article 8 of the ILC Draft Articles though, which aims to link the state with a specific agent) is still relevant for determining the scope of these positive obligations. Public authorities are established to fulfill prescribed aims and they are conferred powers. They assume control over areas of activity, in this way putting themselves in proximate relationship with harm that might arise in these areas.77 It follows that control structures lines of proximity. The more control, the closer proximity may be expected between state conduct and harm, and accordingly, the positive obligations are more demanding. These positive obligations are thus commensurate with the extent of the control. In this sense, we can see some extension of the logic that applies to the rules on attribution. This normative account fits current practice since, as we will see below, it is reflected in the case law under the ECHR. The question which emerges at this junction is what principles apply if the state authorities are out of control. Do states have positive obligations when they do not have control? Can states absolve themselves from responsibility under human rights law by simply saying that they did not have control? Can they decide to relinquish control and free themselves from responsibility? The presumption that operates is that states have control over their territory and therefore continue to be under the obligation to ensure the rights enshrined in the ECHR.78 This implies that States have to re-assert control and take measures to secure these rights as a matter of principle.79 This certainly does not translate into state responsibility for failure to ensure human rights in every concrete case. States might face some practical difficulties in re-asserting control and in the degree of control that they can practically exercise; however, these will have to be taken into account in the assessment of the scope of the positive obligations in the particular case. In addition to issues of practicality and feasibility, certain areas of activities are underpinned with the normative assumption that the state must wield more control. I engage with these areas in more detail in Section 5.D below. First, however, I turn to circumstances where state agents inflict harm and how the positive obligations in relation to this harm are shaped by the degree of control exercised. C. Control and Prevention of State-inflicted Harm Positive obligations are usually analysed in circumstances when private actors cause harm. However, they are as much relevant in circumstances where state agents inflict harm. States are equally under the obligation to structure the relationships between their agents and individuals in such a way that harm is prevented and, if it occurs, is adequately addressed. In this context, the harm caused by failures in these structures is more closely proximate to the state and, accordingly, the obligation to prevent it is more demanding. For these reasons and in relation to the right to life, the Court has observed that ‘[w]hen lethal force is used within a “policing operation” by the authorities, it is difficult to separate the State’s negative obligations under the Convention from its positive obligations’.80 Still, in its analysis the Court distinguishes the two, and to this effect it has been established that Article 2 requires careful scrutiny not only as to whether the use of force by state agents was strictly proportionate to the aim of protecting persons against unlawful violence (negative obligation),81 but also whether the overall operation was ‘planned and controlled by the authorities so as to minimize, to the greatest extent possible, recourse to lethal force’ (positive obligation).82 The intensity of the above positive obligation is influenced by the level of control that the state has over the situation. For example, in Giuliani and Gaggio v Italy, the absence of foreseeability as to the course of the events, as well as the ensuing reduced level of control by the state over the situation, were taken into account in determining whether the organization and the planning of the policing operations were compatible with the obligation to protect life.83 Similarly, in Isayeva v Russia, where the applicant’s close relatives were killed by indiscriminate bombing by the Russian military, one of the factors considered by the Court was that the military operation conducted by Russia was not spontaneous and, therefore, the state had control over the circumstances.84 Likewise, in Makaratzis v Greece the Court did not overlook the fact ‘the applicant was injured during an unplanned operation which gave rise to developments to which the police were called upon to react without prior preparation.’ In the latter judgment it was also added that if the unpredictability of the events and the resultant reduction of the level of control were not taken into account in the assessment of the positive obligations, this might lead to the imposition of an ‘impossible burden on the authorities’.85 In the course of assessing the control over the situation and the scope of the positive obligations, the reasons as to why the state authorities did not have control are also of importance. Absence of foreseeability could be a reason. In Makaratzis v Greece, however, it was established that the degeneration of the situation and the ensuring chaos was largely due to the fact that at the time neither individual police officers nor the chase, seen as a collective police operation, had the benefit of the appropriate structure which should have been provided by the domestic law and practice.86 Since the absence of control was seen as attributable to deficiencies in the national legislation, Greece was found to have failed to protect the right to life. Therefore, despite the unpredictability of the events and the ensuing reduced level of practical control over the circumstances, the normative expectation that police operations are regulated by laws was determinative. Finogenov and Others v Russia, a case involving hostage-taking in a theatre that was stormed by the Russian authorities after the injection of an unknown narcotic gas, strongly supports the argument that there is a correlation between the control and the scope of the positive obligations. In Finogenov and Others v Russia, the Court modified the well-established Osman test, by observing: The authorities’ positive obligations under Article 2 of the Convention are not unqualified: not every presumed threat to life obliges the authorities to take specific measures to avoid the risk. A duty to take specific measures arises only if the authorities knew or ought to have known at the time of the existence of a real and immediate risk to life and if the authorities retained a certain degree of control over the situation.87 The Court delineated a different intensity of obligations and applied a different degree of scrutiny depending on the degree of control that the state had over the situation. It distinguished Finogenov from the above mentioned Isayeva v Russia since ‘[t]he hostage taking came as a surprise for the authorities, so the military preparations for the storming had to be made very quickly and in full secrecy.’ In addition, the authorities were not in control of the situation inside the theatre where the hostages were kept, which was also of material consideration.88 In contrast to the storming of the theatre, however, the positive obligations in relation to the subsequent rescue operation were of different intensity since ‘no serious time constraint existed and the authorities were in control of the situation.’ This justified a different approach when assessing the conduct of the Russian authorities.89 While the use of the gas and the storming was found not to be disproportionate measures in breach of Article 2, the rescue and evacuation operation were found inadequate. The latter operation was subjected to more thorough scrutiny because it was not spontaneous. In addition, it could be expected from the authorities that they had some general emergency plan and some control of the situation outside the building where the rescue efforts took place. The predictability of the harm also implied a higher level of control and more demanding positive obligations: ‘the more predictable a hazard, the greater the obligation to protect against it.’90 In conclusion, control over the situation is crucial for a finding of proximity, and important for assessing the intensity of the positive obligation. At the same time and as already intimated in the end of Section 5.B above, positive obligations themselves might require more pervasive control in certain circumstances. The next section will expand on this argument. Prior to this, however, the circularity in the above argumentation needs to be acknowledged. If states have control they put themselves in closer proximate relations with harm that might materialize and the positive obligations invoked in these circumstances are more robust. At the same time, the positive obligations themselves might require more control by the state. Instead of occluding this paradox, it should be rather openly acknowledged. It is beyond the scope of this article to resolve it because it is part of a much broader issue that ultimately concerns the role of the state in the society and to what extent and under what circumstances this role should be more intrusive.91 D. Assumption of Control in the Area of Public Services A cluster of positive obligations cases involve provision of public services. These services can be provided by public bodies or by private bodies. It has been well-established in the case law that states are not absolved of their human rights law obligations by delegating certain services to private bodies.92 The designation of the body is thus not of significance. For example, the Court has observed that ‘in the public-health sphere, these positive obligations require the State to make regulations compelling hospitals, whether private or public, to adopt appropriate measures for the protection of patients’ lives.’93 The nature of the activity, however, that is, public services, is of important material consideration since the breadth of the positive obligation increases with a function of this nature. The normative assumption that operates in this context is that the state should assume control in relation to these services and, as a consequence, the intensity of the obligation rises. A question which emerges here is how to define public services.94 The importance of the interest at stake, and also the relational context, can be determinative in delineating the contours of this definition. For example, in relation to education, the Court has emphasized that children have no alternative but to attend school since primary education is obligatory.95 In O’Keeffe v Ireland, the Court went so far as to establish an ‘inherent obligation [of the State] to protect children in this context, of potential risks.’96 The applicant in this case complained that the system of primary education failed to protect her from sexual abuse by a teacher. In relation to this complaint, the Court held that the primary education context of the present case defines to a large extent the nature and the importance of this obligation. The Court’s case law makes it clear that the positive obligation of protection assumes particular importance in the context of the provision of an important public service such as primary education, school authorities being obliged to protect the health and well-being of pupils and, in particular, of young children who are especially vulnerable and are under the exclusive control of those authorities.97 Although the school in question was owned and managed by a non-state actor, the Court relied on the fact that education is an importance public service and on the vulnerability of children to determine the scope of the positive obligation.98 The importance of the activity in question and the control of the authorities were also emphasized in Ilbeyi Kemaloglu and Meriye Kemaloglu v Turkey, a case about the child who froze to death after being left alone by the school authorities in a heavy snow storm: [T]he State’s duty to safeguard the right to life is also applicable to school authorities, who carry an obligation to protect the health and well-being of pupils, in particular young children who are especially vulnerable and are under the exclusive control of the authorities.99 The same principle has been applied to psychiatric institutions, in relation to which ‘the State remained under a duty to exercise supervision and control’100 and nursing homes, which the state is under the obligation to regulate.101 Similar reasoning underpinned Iliya Petrov v Bulgaria, where a boy was severely harmed after entering a transformer and receiving an electric shock. The Court acknowledged that the boy was very unwise to enter such a dangerous place, but it still held that the ‘decisive factor’ for the occurrence of the incident was the deficient control by the authorities regarding the safety of electric transformers,102 the latter being a public service in relation to which the state is expected to exercise control. Cevriogli v Turkey is also illustrative in this respect. The case concerns the death of a boy as a result of falling into a large water-filled hole outside a private building under construction in a residential area. The Court referred to the ‘inherently hazardous nature’ of construction sites and, accordingly, the expectation that the state controls, inspects and supervises the activities at these sites.103 In its reasoning, the Court added that the state ‘in the present context had a more compelling responsibility towards the members of the public who had to live with the very real dangers posed by construction work on their doorsteps.’104 The above outlined expectation that in certain areas the state assumes control, shapes the approach to causation. For example, in Cevriogli v Turkey, it was admitted that no causal link may exist between the failings to inspect the construction site and the death of the boy ‘for the purposes of civil liability.’ However, in the context of state responsibility where the objective is to find the responsibility of a collective, strict causation lines are inappropriate. In this respect, the Court observed that ‘proper implementation of an inspection mechanism would undoubtedly have increased the possibility of identifying and remedying the failings which were responsible for the death of the applicant’s son.’105 E. Source of the Harm and the Related Level of Control The establishment of causation between failures on behalf of the state to take measures and harm is also affected by the source of the harm: whether the source is a natural or a man-made phenomenon. In both contexts, the state is under the general obligation to protect; however, in the event of a harm ensuing from a man-made phenomenon, proximity is easier to establish and the scope of the positive obligations is more demanding. In situations involving ‘dangerous activities’, where the harm is perceived as man-made, or in relation to events ‘regulated and controlled by the State’, it is easier to establish that state omissions are causative to harm. An example to this effect is Öneryildiz v Turkey, a case about an explosion at a garbage collection point. In its submissions to the Court, Turkey tried to challenge the extension of positive obligations under Article 2 to all circumstances of unintentional death. The Court, however, emphasized the dangerous nature of the activity and the ensuing expectation that the state regulates it. In this way, it established a proximity between the harm and failure by the state to regulate,106 which also implied that the positive obligation to ensure the right to life ‘must be construed as applying in the context of any activity, whether public or not, in which the right to life may be at stake, and a fortiori in the case of industrial activities, which by their very nature are dangerous, such as the operation of waste-collection sites.’107 In Vilnes and Others v Norway, the Court extended this logic to risky activities and observed that ‘it sees no need to consider in detail the degree of involvement of the respondent State in the hazardous activity in question, since the Convention obligation applies “any activity, whether public or not”’.108 The same approach was adopted in Kolyadenko and Others v Russia, a case about an urgent massive evacuation of water from reservoir, where the Court also emphasized that a reservoir is a man-made industrial facility.109 The above analysis is not modified by the nature of the agent that performs the activity. Whether that activity is performed by a public entity or by, for example, private corporations, is immaterial. Thus, the Court’s assertion in Brincat and Others v Malta, that the positive obligation to safeguard lives ‘may apply in cases, such as the present one, dealing with exposure to asbestos at a workplace which was run by a public corporation owned and controlled by the Government’ is confusing.110 If consistency in the case law is to be maintained, the public or private nature of the entity that engages in the activity is not pertinent. As the Court itself has observed, the positive obligation to protect life applies ‘in the context of any activity, whether public or not’.111 The above cases implicating man-made harm can be distinguished from circumstances where harm is caused by ‘natural’ disasters. It is more difficult to establish proximity and therefore the positive obligations are not that extensive in cases of natural disasters ‘which are as such beyond human control’ and ‘do not call for the same extent of State involvement’.112 In Budayeva and Others v Russia, a case about a mudslide causing loss of life and destruction of property, the Court observed that ‘[t]he scope of the positive obligations imputable to the State in the particular circumstances would depend on the origin of the threat and the extent to which one or the other risk is susceptible to mitigation.’113 It was added that the consideration of not imposing a disproportionate burden on the state ‘must be afforded even greater weight in the sphere of emergency relief in relation to a meteorological event, which is as such beyond human control, than in the sphere of dangerous activities of a man-made nature.’114 Still, against the background of the national authorities’ omissions in the implementation of land-planning and emergency relief policies in the hazardous area where the mudslide occurred, and the existence of a causal link between these failures and death of the victim, Russia was found to have failed to discharge its positive obligations under Article 2.115 The same logic has been extended in relation to the positive obligations corresponding to the right to property. In Öneryildiz v Turkey, the Court observed: Genuine, effective exercise of the right protected by that provision [Article 1, Protocol 1] does not depend merely on the State’s duty not to interfere, but may require positive measures of protection, particularly where there is a direct link between the measures an applicant may legitimately expect from the authorities and his effective enjoyment of his possessions.116 The establishment of this causal link is affected by the source of the harm. In Öneryildiz v Turkey such a link was found to be present: ‘there is no doubt that the causal link established between the gross negligence attributable to the State and the loss of human lives also applies to the engulfment of the applicant’s house.’117 In Budayeva and Others v Russia, however, the Court introduced a distinction along the following lines. As opposed to Öneryildiz v Turkey, where the events occurred ‘under the responsibility of the public authorities’ and where ‘[t]reatment of waste … is regulated and controlled by the State, which brings accidents in the sphere within its responsibility’,118 in Budayeva and Others v Russia the Court considers that natural disasters, which are as such beyond human control, do not call for the same extent of State involvement. Accordingly, its positive obligations as regards the protection of property from weather hazards do not necessarily extend as far in the sphere of dangerous activities of a man-made nature.119 While the implementation of certain measures (mud-defence infrastructure and an early warning system) were considered by the Court as vital for the protection of life in case of a mudslide (and in this sense the Court was prepared to find causality between the absence of these measures and the loss of life), no evidence was found that these measures could have prevented the damage to the applicants’ possession: ‘[I]t cannot be said that the causal link between the State’s failure to take these measures and the extent of the material damage is similarly well-established.’120 In Budayeva and Others v Russia, the Court thus appears to suggest that to a certain extent some of the damage to property was imputable to state negligence; however, it is difficult to ascertain how much. The Court’s approach is also based on the premise that even if mud-defence infrastructure had been installed, damage to property would still have occurred. F. Assumption of Control over the Victim121 While the preceding sections addressed the issue of state control over certain circumstances and activities, this section focuses on control over individuals. Once the state has undertaken any special responsibilities in relation to certain individuals, the lines of proximity between harm and state omissions solidify. States owe more extensive positive obligations to those with whom they have special ties. The primary example in this respect are prisoners, who are placed under extensive state control.122 This control implies more demanding positive obligations. These have been considered in various contexts (protection from private violence,123 protection from suicide,124 acceptable detention conditions, including prevention of health hazards,125 and provision of minimum socio-economic assistance). It will suffice to compare the scope of the positive obligations in relation to the provision of health care for prisoners with the scope of the positive obligation in relation to provision of healthcare to the population at large.126 In the latter context, the Court applies a high definitional threshold to find violations of Articles 3 and 8 in the sphere of socio-economic assistance.127 The Court has not excluded the possibility that violations of Articles 3 and 8 might be found in circumstances characterized with deprivations resulting from, for example, insufficient welfare benefits; however, such a finding will be made under exceptional circumstances.128 The exceptionality approach, however, will be modified if other elements intervene. As mentioned above, no such approach is applied in relation to prisoners. The Court has taken this very far since not even financial considerations (for example, the argument that the state does not have enough money to maintain prisons) are accepted.129 There are other circumstances where, due to the special relationship between the state and the victim and more specifically due to the assumption of control by the state over the victim, the logic of exceptionality is displaced. In Denis Vasilyev v Russia, the Court observed that the duty to protect is not ‘confined to the specific context of the military and penitentiary facilities.’130 It added: It also becomes relevant in other situations in which the physical well-being of individuals is dependent, to a decisive extent, on the actions by the authorities, who are legally required to take measures within the scope of their powers which might have been necessary to avoid the risk of damage to life and limb.131 The factual substratum of Denis Vasilyev v Russia was underpinned by omissions by police officers. In particular, after finding the applicant unconscious on the street, they left without calling for medical assistance. Despite the important assertion framed by the Court in the above quotation to the effect that the obligation to protect is triggered when the well-being of the individual is dependent on the state, the reasoning in the judgment overall is confusing. The Court pointed to various factors (vulnerability of the person, knowledge about his position, control over him once the state authorities knew about his position and the requirements under the national legislation to render assistance) to find that the state was in breach of its positive obligations. Overall, the reasoning manifests a heavy emphasis on the requisites of the national legislation to assist the person, which tilted the Court to find a violation of Article 3. In the next section, I will discuss in more detail the role of domestic legality. Here it is pertinent to observe that there might be other situations where the above mentioned exceptionality might be displaced. One such example is where the state has precluded the availability of alternative means of protection and assistance. Such a situation might transpire in relation to asylum-seekers, who might not be allowed to work in the first place, which will inevitably modify the analysis as to whether the absence of socio-economic assistance by the state causes harm falling within the scope of Article 3 or 8.132 Because of the special relationship between the person and the state that arises in the above-mentioned contexts, the foreseeability of the harm requirement might be loosened.133 In light of the special position of the victim in relation to the state, it might be hardly possible to argue that the state did not know or ought not to have known about the harm or the risk of harm. Some aspects of the case law, however, cause confusion. For example, in Nencheva v Bulgaria, the Court extensively reviewed whether the central national authorities knew about the dire circumstances of the disabled children who were accommodated in a home. In light of the fact that the home in question was under the control of these authorities, one is left to wonder whether the authorities should in any case have known about the risks faced by the children. Even if they did not actually know, there should have been mechanisms for channeling such information.134 In sum, the case law is ambiguous as to which factor has a dominating role: the actual knowledge about harm, on the one hand, or the normative supposition that state authorities should know or necessarily knew about the harm (or risk of harm) to individuals under their control. One final observation is due in this section. The special position of the victim can be related to the particular vulnerability of certain categories of persons.135 The Court has recognized various groups as vulnerable: children,136 asylum-seekers,137 prisoners, military conscripts,138 persons with disabilities,139 victims of domestic violence140 and Roma.141 It is useful, however, to distinguish between the sources of these various vulnerabilities. Some of them might be innate and inherent vulnerabilities (for example, children). Others might be related to the social context (for example, Roma, religious minorities).142 In this section, I draw attention to a distinctive vulnerability that stems from the specific relationship with the state and, in particular, from the exposure of the person to state power. Prisoners, for example, who due to their detention are exposed to state power and deprived of other sources of help, are placed in a special relationship with the state, which explains the more demanding positive obligations.143 Certainly, this special relationship could be intimately related to or substantiated by vulnerability stemming from innateness or social context (for example, protection of children in the context of compulsory education) and in this sense, the different sources of vulnerability might be interrelated. 6. Techniques for Avoiding Causation So far this article has covered areas where traditionally the state has had an important role and in this sense the state more easily places itself in closer proximate relations with harm that might materialize (for example, policing operations, provision of public services, industrial activities, restraints imposed upon individuals). In other areas, however, the intensity of the involvement of the state can be more controversial (for example, protection of the environment, regulation of private companies) and, as already suggested at the end of Section 5.C, the question as to how intrusive the role of the state should be and how much control the state should exercise can be more contentious. Moreover, the establishment of causation between harm and state omissions might be hampered by empirical and epistemological uncertainties. Since the Court might not want to see itself deeply implicated in normative judgments about the role of the state in society more generally, and since the Court might not be in a position to resolve empirical and epistemological uncertainties, it has crafted techniques to avoid making these judgments and resolutions. These techniques also imply that the Court does not have to directly confront the issue as to whether state omission is causative to harm. Two such techniques will be discussed here: domestic legality and procedural protection. A. Domestic Legality When an omission is contrary to the national regulatory framework, proximity between this omission and the harm amounting to violation of human rights law is easier to assume. In other words, when the national legislation or applicable regulatory standards themselves envisioned the undertaking of certain measures and these were not performed, the Court is more willing to accept that there is a nexus between the non-performance and harm. The underlying assumption is that the national regulatory framework was adopted in order to prevent harm. Once this is transposed at the level of the ECHR, whether or not this is indeed the case (to wit, whether or not generally or in relation to the particular applicant the proper application of the legal framework would have indeed prevented harm or reduced the risk of harm), seems to be less relevant, since the above assumption continues to operate. How non-compliance with the national legal requirements renders the proximity standard less stringent from the perspective of the applicant was made obvious in I v Finland. The applicant complained that a hospital has failed to guarantee the security of her data against unauthorized access. She worked as a nurse and was diagnosed as HIV-positive. At a certain point, she suspected that her colleagues were aware of her illness and soon her contract was not renewed. On the facts, it was not possible to determine whether her records were actually accessed by an unauthorized third person. As a consequence, it was not possible to prove a causal connection between deficiencies in the access rules in this particular hospital (that is, not maintaining a log of all persons who had accessed her medical files) and the harm that she has experienced (that is, dissemination of information about her medical condition). However, the Court did not find it necessary to prove such causation. Instead, it observed that ‘what is decisive is that the records system in place in the hospital was clearly not in accordance with the legal requirements’ in the national legislation and did not hesitate to find a failure on behalf of the state to fulfill its positive obligation under Article 8.144 The requirements laid down in the national legislation, including those that require positive measures, establish a baseline. Any deviation from this baseline is suspect and makes it easier to argue that a state’s failure to comply with its own baseline is causative of harm. Support for this approach can be found in the context of the right to life and the adequate standard of health care.145Lopes de Sousa Fernandes v Portugal and Elena Cojocaru v Romania are judgments in point. The Court observed that it would not speculate on the particular patient’s prospects for survival, if the measures required by the medical protocols had been actually undertaken. It sufficed for the establishment of state responsibility that there was ‘apparent lack of coordination of the medical services and … delay in administering the appropriate emergency treatment,’ which ‘attest to a dysfunctionality of the public hospital services’.146 In the reasoning of the majority, this dysfunctionality could be related to non-observance of medical protocols. On the facts, however, it was not entirely clear whether non-compliance with the protocols caused the death of the applicants’ relatives. The Court’s leniency regarding causality in the above-mentioned two cases prompted Judges Sajó and Tsotsoria to attach a Dissenting Opinion to the Chamber judgment in Lopes de Sousa Fernandes, in which they observed that it is hard to understand ‘how an alleged organisational negligence that did not result in death can be construed as the basis of State responsibility for failing to protect life.’ In their Joint Dissenting Opinion, they also added that even if there had been a causal relation, this is still not enough to find a violation. I will return to this aspect of their argument in Section 7.C below, where I address the distinction between incidental and systemic failures. Here is it pertinent to observe that the Grand Chamber in Lopes de Sousa Fernandes addressed the dissenters' concerns by observing that the question whether there has been a failure by the State in its regulatory duties calls for a concrete assessment of the alleged deficiencies rather than an abstract one. … Therefore, the mere fact that the regulatory framework may be deficient in some respect is not sufficient in itself to raise an issue under Article 2 of the Convention. It must be shown to have operated to the patient's detriment.147 It follows that the Grand Chamber has modified the Chamber's approach. As the above quotation shows, causality needs to be established between the organisational negligence and the condition of the specific applicant. Using the national regulatory framework as a metric has been applied in other areas of the case law. More specifically, in A v Croatia, a domestic violence case, the failure by the national authorities to implement measures ordered by the national courts was highlighted, and led the Court to conclude that the respondent state failed to ensure the victim’s right to private life.148 In Taskin and Others v Turkey, the applicant complained that the operation of a gold mine posed risks to their right to life and private life. The respondent government challenged the assertion that the operation of the mine had harmful effects, and in this sense scientific uncertainty permeated the facts. The Court did not find it necessary to engage with the issue as to whether the operation of the mine was indeed contributory to harm, since this operation was contrary to domestic law.149 Certainly, the state might have perfectly complied with the existing national legal and regulatory framework and still have failed to fulfill its positive obligations under the ECHR, since deficiencies in this very framework might be causative to harm.150 On the other hand, even if a failure to take certain protective measures was contrary to the national law and regulation, this is not in itself conclusive that the state has failed to fulfill its positive obligations under the ECHR. The Court might pursue further enquiries into lines of proximity.151 Domestic legality is thus not a conclusive test.152 This renders the analysis distinctive in comparison with cases framed as implicating negative obligations, where if a measure restricting the right is not ‘in accordance with the law’, this automatically renders the restriction contrary to human rights law.153 Still, as clarified above, even in the context of positive obligations, domestic legality plays an important role in shaping lines of proximity. In addition, the notion of legality does not only have a substantive aspect, which, as explicated above, concerns compliance with domestic regulatory frameworks; it also has a procedural aspect, to which I now turn. B. Procedural Protection In various fields of its case law on positive obligations, the ECtHR has added a procedural layer to the scope of the Convention rights by requiring states to ensure the availability of effective national procedures. As Brems and Lavrysen have observed, the rationale behind this move is the proposition that procedural guarantees are instrumental for better protection of the substantive guarantees.154 The task of this section is not to offer a general analysis of the development of procedural protection by the Court, which has been already done by the above-mentioned authors. Its focus is rather the underlying causality, namely the understanding that the substantive harm is less likely to have materialized, had the decision-making process at national level been of sufficient quality. The main argument that I advance here is that procedural protection offers an avenue for the Court to deal (or rather not to deal) with empirical and epistemological uncertainty. This uncertainty poses challenges in ascertaining the remoteness or the proximity between the harm sustained by the individual and state omissions; procedural protection resolves the difficulty. The Court can eschew conclusive determinations that certain substantive omissions cause harm, and instead can focus on procedural omissions and deficiencies at national level. The inclusion of a right to access to information in the environment-related judgments delivered by the Court supports the above argument.155 In Roche v United Kingdom, there was uncertainty whether the applicant had been put at risk through his participation in chemical tests. This uncertainty did not have to be resolved because the Court framed its task as considering whether a positive obligation arose to provide an “effective and accessible procedure” enabling the applicant to have access to “all relevant and appropriate information” which would allow him to assess any risk to which he had been exposed during his participation in the tests.156 The Court also added that the applicant’s uncertainty ‘as to whether or not he had been put at risk through his participation in the test carried out at Porton Down could reasonably be accepted to have caused him substantial anxiety and stress.’157 Accordingly, the core issue in the case was shaped as not how the tests themselves contributed to the harm falling within the scope of Article 8; but how the denial of access to information about the tests caused harm falling within the scope of Article 8. These are certainly two distinct, although related, types of harm. Access to information is only one element for assessing procedural protection. As Hatton and Others v United Kingdom shows, the Court can scrutinize more generally the quality of the national decision making process.158 In particular, scientific uncertainty underpinned the case since ‘[t]he position concerning research into sleep disturbance and night flights is far from static’.159 No finding on this specific point had to be made at the level of the Strasbourg Court because the national decision-making process was found adequate. No fundamental procedural flaws in the preparation of the night flight scheme at Heathrow airport were found. Vilnes and Others v Norway is an example of a case where quite wide-ranging assumptions about causal relationships and procedural protection were made.160 This merits more detailed examination. The applicants, who worked as divers in the North Sea for private companies, complained that Norway did not adopt an effective legal framework of safety regulations that could prevent the divers’ lives and health being put at risk. While dismissing most of the divers’ allegations, the ECtHR still found a failure on behalf of Norway. The underlying reason was that the companies were left with little accountability vis-à-vis the state authorities in relation to the usage of diving tables, which were treated as company business secrets.161 In other words, Norway allowed a situation in which the divers were not informed about the health and life-related risks pertaining to the usage of diving tables. There are two types of causality underpinning Vilnes and Others v Norway that have to be separated and further clarified. The first relates to the extent to which the rapid decompression tables did in fact contribute to the applicants’ medical problems. The standard for assessing this contribution was formulated in the following way by the Court: The Court, having regard to the parties’ arguments in the light of the material submitted, finds a strong likelihood that the applicants’ health had significantly deteriorated as a result of decompression sickness, amongst other factors. This state of affairs had presumably been caused by the use of too-rapid decompression tables. … Thus, with the hindsight at least, it seems probable that had the authorities intervened to forestall the use of rapid decompression tables earlier, they would have succeeded in removing more rapidly what appears to have been a major cause of excessive risk to the applicants’ safety and health in the present case.162 However, as the Court framed the case, the core issue was not that the state had not prevented the use of rapid decompression tables as such or that the state had not prevented their use earlier. As the Court alluded in the above citation, the conclusion that elimination of the tables would have reduced the risk can be reached only with the benefit of hindsight. At the material time in the past, it was widely believed that diving did not have serious long-term effects in the absence of decompression sickness. A procedural duty was rather placed at the heart of the case, namely the duty of the state to provide information essential for the divers to assess the health risks. Although the risks associated with diving at that time were still disputed, the raison d’etre of the decompression tables themselves was to provide information essential for the assessment of risk to personal health.163 As the Court observed, the authorities were aware that the diving companies kept the tables confidential for competitive reasons.164 The authorities failed to enlighten the divers about the risks, which in the logic of the judgment would have enabled the divers to give informed consent to the taking of such risks.165 In light of the above, a second type of causality was brought forward in the judgment: if the state had ensured that the divers could assess the risks to their health and give informed consent to the risks involved, this would lead to the elimination of the use of the rapid tables. The Court frames this in the following way: Had they done so [had the authorities ensured that the companies provide information by not keeping the tables confidential] they might conceivably have helped to eliminate sooner the use of rapid tables as a means for companies to promote their own commercial interest, potentially adding to the risks to divers’ health and safety.166 Two comments are pertinent here. First, the standard of ‘conceivably have helped’ as a test of causation appears to be very low, and has not been used before Vilnes and Others v Norway. Secondly, the reasoning in the above quotation leads to contradictions within the judgment. In light of the absence of clear expert understanding at the material time of the consequences of the usage of decompression tables, as in fact acknowledged by the Court itself, it is hard to maintain that if the divers themselves had had access to the tables, this information would have been of use and led them to change their behaviour. The above analysis is not a suggestion that the state was not at fault. The wrong that can be clearly imputed to the state is that it allowed the companies to treat the decompression tables as their business secret and keep them confidential for competitive reasons. How that wrong can be linked with the harm sustained by the particular applicants so that this harm can be translated into international responsibility under the ECHR, however, is a separate question. In its response to this question, the Court appeared to be very flexible when assessing causation. Admittedly, this could be counterbalanced with the understanding that it is not reasonable to allow business secrets in relation to issues that raise controversies as to their impact on human health. In other words, it would not have been anything close to an unreasonable burden on the state to demand disclosure of the tables. It is precisely the test of reasonableness that the next section will address. 7. Technique for Limiting Responsibility when Causation is Present A. Reasonableness The Court has consistently reiterated that ‘the positive obligation to protect is to be interpreted in such a way as not to impose an excessive burden on the authorities.’167 The scope of the positive obligations under the Convention cannot therefore be unreasonable. When the Court refers to reasonableness in the context of positive obligations, it has in mind public interests—including public policy considerations, budgetary concerns and the rights of others—as factors that might compete with the assistance and protection interests of the particular applicant.168 This section considers how the reasonableness standard might affect and mould the assessment of causation. As already suggested above in relation to Vilnes and Others v Norway, even if the lines of causation between state omissions and harm are tenuous, reasonableness might intervene and buttress a finding of a violation. Here I will focus on the other end of the spectrum, where even if there is clear factual causality between state conduct and harm, additional factors related to the reasonableness standard influence the determination whether the respondent state is responsible under human rights law. Mastromatteo v Italy is illustrative in this respect. A brief summary of the facts is apposite here. The applicant’s son was murdered by a gang of criminals. The murder was carried out at a time when they were on special prison leave or benefiting from a regime of semi-liberty. It was undisputed that if the state had not released the criminals, Mastromatteo would not have been murdered. In this sense, one can see a clear causation: failure by the state to keep them in prison resulted in severe harm. The Court, however, observed that a mere condition sine qua non does not suffice to engage the responsibility of the State under the Convention; it must be shown that the death of A. Mastromatteo resulted from a failure on the part of the national authorities to ‘do all that could reasonably be expected of them to avoid a real and immediate risk to life of which they had or ought to have had knowledge’, the relevant risk in the present case being a risk to life for members of the public at large rather than for one or more identified individuals.169 The Court went on to examine the decision of the national authorities to let the criminals on leave and concluded that there was nothing in the material before the national authorities to alert them to the fact that the release of M.R. or G.M. would pose a real and immediate threat to life, still less that it would lead to the tragic death of A. Mastromatteo as a result of the chance sequence of events which occurred in the present case.170 In light of the benefits associated with rehabilitation programs for prisoners, the judgment suggests that it would not have been reasonable to keep M.R. and G.M. in prison when the state did not know that they posed ‘real and immediate’ risk to harm. The Court thus took note of broader considerations related to reasonableness and the interest of others,171 namely the benefits associated with letting prisoners on leave for the purpose of social reintegration. B. The ‘Real and Immediate Risk’ Standard A question that merits further investigation at this junction relates to the role of the ‘real and immediate risk’ test and, more specifically, how it affects causation and the reasonableness standard. In support of the standards set in the above-quoted paragraphs from Mastromatteo v Italy, the Court referred to Osman v United Kingdom. The test of ‘real and immediate risk to life’ was established therein in the context of the positive obligation to take protective operational measures. For a violation of this obligation to be found it must be established to its [the Court’s] satisfaction 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 or individuals from the criminal acts of a third party and that they failed to take measures within the scope of their powers which, judged reasonably, might have been expected to avoid that risk.172 This test is relevant in circumstances where the state knew or should have known about dangers for a specific identifiable individual from the criminal acts of another individual.173 The Court has never clarified the meaning of ‘real’ and ‘immediate’ and usually bundles these criteria together without elucidating how they differ. Often, it refers to them as ‘general principles’ without clarifying their application to the concrete circumstances of the case. ‘Real’ risk could be understood as risk that is objectively given and not merely speculative. The qualifier ‘real’ could refer to the likelihood that the risk will materialise.174 In this sense, a parallel can be drawn with the ‘real prospect’ test discussed in Section 4, which refers to the likelihood that a protective measures could obviate the risk. ‘Immediate risk’ could be understood as risk that is ‘present and continuing’.175 It has also been suggested that immediacy implies that the harm could be expected to ‘materialize at any time’.176 The immediacy prong of the test and the identifiability of the person exposed to the immediate harm are the distinguishing features that characterize the circumstances where the positive obligation of taking protective operational measures is applicable. The existence of a ‘real and immediate risk’ needs to be conceptually separated from the existence of official knowledge about the risk (the latter requirement was addressed in Section 3 above). Such a risk might exist even if the state authorities are not aware of it, and thus state responsibility will be precluded when there is no official awareness. However, this preclusion of responsibility is undermined by the requirement formulated as ‘ought to have known’ about the risk, which is given as an alternative to ‘the authorities knew’ standard. As already observed in Section 3 above, the ‘ought to have known’ test implies calculations as to how reasonable it is to actually expect that the authorities know about and foresee harm. It is important to clarify here that the ‘real and immediate risk’ test is not of general applicability to positive obligations. Accordingly, McBride’s argument that the Court has responded to the difficulties arising from different possible sources of harm by ‘adopting an exacting causal test, namely that the risk of death from the supposed threat be “real and immediate”’ needs to be rejected.177 This test is rather limited to circumstances where the person exposed to the harm can be individualized and identified in advance, and the risk of harm is immediate. In this sense, Osman v United Kingdom and the other cases in which the positive obligation of taking protective operational measures has been found to be triggered,178 have a factual substratum that only tenuously resembles Mastromatteo v Italy. The ‘real and immediate risk’ test might appear well-suited in cases where identifiable individuals are exposed to the risk of criminal acts of others, since it might be unreasonable to expect the authorities to take protective operation measures of an ad hoc nature without some imminence and concreteness of the risk to which a particular individual is exposed. In other words, it would be unreasonable to expect the state to take such measures in relation to an indeterminate class of people. In contrast, when the issue is general protection of the society, the test of ‘real and immediate risk’ is ill-suited. Therefore, the factual substratum of Mastromatteo v Italy does not call for the invocation of protective operational measures since there was no immediate danger to the life of any specific individual; it rather reveals circumstances where the applicant’s son happened to be an accidental victim of various errors in relation to the application of the national regulations about prison leave.179 These errors were, however, assessed as not sufficiently material and systemic, an issue I address in the next subsection. In sum, ‘the real and immediate risk’ test limits the circumstances where the state might be found responsible for its omissions. However, this limitative function is not of general applicability.180 It is not appropriate to operate in circumstances demanding general protection of the society, where there is no immediate danger to any specific individual,181 and where the state is rather under the obligation to put in place a general legislative and administrative framework for regulating activities so that harm is prevented.182 The current uncertainty as to the meaning of ‘immediate’ risk, however, and the threshold of immediacy required, will continue to spread confusion in the case law.183 The reason for this confusion is not simply the Court’s inadequate stringency in the application of the standards, a point already intimated above; there are also broader considerations that need to be seriously considered. In particular, if the state were to take protective actions against any potential risk regardless of its immediacy, then we might be confronted with the problem of a too intrusive state. This is a dilemma already identified in Section 5 above and, it calls for caution when expanding the scope of the positive obligation of putting in place general legislative and administrative framework for regulating activities so that harm is prevented. On a related point, we as a society might have to accept certain levels of risk,184 and this might militate against expansive construction of positive obligations. This is an argument that bulwarks the proposition that even in cases of clear factual causality, violations of human rights should not be found because society has to tolerate and accept certain risks. C. Incidental as Opposed to Systemic Failures In addition to the ‘real and immediate risk’ test, another technique utilized by the Court for limiting state responsibility relates to the distinction between incidental and systemic failures. As touched on above, a reason that might explain the outcome in Mastromatteo v Italy, where no violation was eventually found, is that the errors on the part of the national authorities appeared incidental and not systematic.185 In this sense, it might be unreasonable to find Italy responsible under ECHR for isolated errors and omissions even if they might have caused harm. This line of reasoning clearly emerges in the context of Article 2 and health care: [W]here a Contacting State had made adequate provision to secure high professional standards among health professionals and to protect the lives of patients, it cannot accept that matters such as error of judgment on the part of a health professional or negligent coordination among health professionals in the treatment of a particular patient are sufficient of themselves to call a Contracting State to account from the standpoint of its positive obligations under Article 2 of the Convention to protect life.186 As pointed out in the Joint Dissenting Opinion of the Chamber judgment in Lopes De Sousa Fernandes v Portugal, the majority had set a standard that casual acts of negligence by members of staff would not give rise to a substantive breach of Article 2. The Grand Chamber judgment in Lopes De Sousa Fernandes v Portugal, not only unequivocally upheld the principle embodied in the above quotation, but it added an additional layer of restrictiveness as compared to the approach by the Chamber in the same case.187 Any deficiencies that might give rise to a substantive violation of Article 2 in medical cases, need not only be systemic or structural (and thus not ‘a mere error or medical negligence’), but they must also implicate denial of immediate emergency care.188 The suggestion that incidental failures might not afford a basis for state responsibility is reminiscent of another test that has been invoked in the case law, namely the ‘significant flaw’ test. The Chamber invoked this test in precluding responsibility in Söderman v Sweden. In particular, it observed that ‘only significant flaws in legislation and practice, and their application, would amount to a breach of a State’s positive obligations under the said provision [Article 8]’.189 Subsequently, however, when the Grand Chamber reviewed Söderman v Sweden, it partially rejected the ‘significant flaws’ test: [S]uch a significant-flaw test, while understandable in the context of investigations, has no meaningful role in an assessment as to whether the respondent State had in place an adequate legal framework [emphasis in the original] in compliance with its positive obligations under Article 8 of the Convention since the issue before the Court concerns the question whether the law afforded an acceptable level of protection to the applicant in the circumstances.190 ‘Acceptable level of protection’ is a standard the Court had never used before.191 What is also interesting about Söderman v Sweden is that no reasonableness standard was even invoked: the Grand Chamber was simply not satisfied that the relevant Swedish law, both criminal and civil, as it stood at the time when the applicant’s stepfather covertly attempted to film her naked in their bathroom for a sexual purpose, ensured protection of her right to respect for her private life.192 In particular, the act of filming was not criminalized. The approach by the Grand Chamber in Söderman v Sweden barely squares with the approach in Mastromatteo v Italy, or in fact, in other judgments where the issue was whether the national legal and administrative framework was effective and where reasonableness was applied as a factor in the delimitation of the scope of the positive obligation.193 Can any explanations concerning this inconsistency be found? An overview of the case law shows that when criminal legislation is invoked as a means of ensuring the rights,194 no test of reasonableness is applied and, in fact, the issue of causation seems to be immaterial. As a consequence, it is simply assumed that criminalization (or an interpretation of the national criminal law so that its reach is more expansive) contributes to better protection of human rights.195 In contrast, when legal frameworks other than criminal law are at issue, reasonableness and competing interests are included as factors in the analysis. Mastromatteo v Italy exemplifies this since ultimately the issue in this case was whether the national framework regulating prison leave contained sufficient safeguards to protect the general population from prisoners on leave. The approach described above, under which the test of reasonableness is not applied when the issue is whether the national criminal law ensures effective protection, is, however, balanced in the following way. Criminalization, as a means of ensuring human rights, is required only where the harm sustained by the victim meets a certain threshold of severity. This will be the case, for example, where Articles 2, 3 or 4 are found applicable,196 and where ‘fundamental values and essential aspects of private life are at stake.’197 In sum, it was relatively easy to reject the ‘significant flaw’ test in Söderman v Sweden because the effectiveness of the criminal law lay at the core of the case. In contrast, the health care cases seem to raise more challenging issues related to allocation of resources and medical expertise. These cases might prompt the Court to be more cautious in finding a substantive violation of Article 2, since inter alia it might be difficult to assess the causal connections between the alleged inappropriate medical treatment and the harm sustained by the specific person. Perhaps for this reason, many of the cases in this area conclude only with a finding of a procedural violation (that is, failure to set up an independent judicial system so that the cause of death of patients in the care of the medical profession can be determined and those responsible held accountable).198 In this way and as already clarified in Section 6.B above, the Court avoids engagement with difficult questions of causation. The difficulties in establishing causal connections might also invite the Court to maintain the distinction between systemic and incidental failures. When confronted with information about deficiencies of a more systemic nature, the Court might more readily link the concrete case with these general shortcomings. 7. Conclusion No hard-edged legal tests apply to cases invoking positive obligations under the ECHR. This flexibility is similarly applied to the requirement for causation, that is, the linkage between the harm sustained by the individual applicant and state omissions. While certainty is not required that the interposition of a missing action would have prevented the harm, no general threshold has been articulated as to how likely it is that a protective measure would have averted the harm. The Court also merges issues of knowledge, reasonableness and causation in its assessment of state responsibility in the framework of positive obligations. Human rights law is thus far from rigid in the assessment of the linkage between state omissions and harm, an approach that can be understood in light of the objective of this body of law, namely assessing the responsibility of a collective (that is, the state). This assessment is underpinned by the assumption that the state is the entity tasked to ensure the rights of the individuals within its jurisdiction. As a consequence, the ECtHR’s approach wavers between effective protection of human rights, on the one hand, and not imposing an unreasonable burden on the state on the other. The establishment of causation is influenced by these considerations. Still, it would not be satisfactory to simply say that the standard of causation applied oscillates between effectiveness and reasonableness. Analytical rigour demands that we further scrutinize the role of causation in the context of positive obligations. This scrutiny shows that by assuming control over certain activities, the state places itself in proximate relationships with harm that might arise in relation to these activities even if this harm is not directly attributable to state agents. Control thus implies closer proximity and more demanding positive obligations. Paradoxically, in certain areas the absence of sufficient control by the state creates the basis for the finding that the state has failed to fulfill positive obligations. This paradox is perhaps only apparent since these areas are underpinned by the normative assumption that the state should assume control. This shapes the approach to causation by making it less stringent. Establishing causation between harm and state omission may be fraught with factual and epistemological uncertainty. In these circumstances, a conclusive determination that the nexus between state omission and harm is too attenuated or sufficiently solid to sustain a violation might be eschewed. Instead, we might rather ask whether the omission was contrary to the applicable domestic legal framework. In cases of non-compliance with this framework, the Court is more prepared to find that the omission has led to harm. Another avenue for avoiding issues of causality is by focusing on the process at national level leading to a decision that is allegedly contrary to states’ positive obligations. If this process is of sufficient quality, the finding of a violation is less likely. The assessment of the quality of the process also includes the availability of procedural guarantees. If these are incorporated at national level, the finding of a violation is also less likely. Finally, I have elaborated on three techniques that might limit the finding that the state is responsible. The standard of reasonableness, and the requirement that the risk of harm is ‘real and immediate’, might have such limiting functions. Although still with uncertain contours, the distinction between incidental and systemic failures might also be determinative. As a consequence, even if a failure is causative to harm, when the concrete case is representative of a mere incidental failure, no state responsibility might be found. ACKNOWLEDGMENTS The author would like to thank Matthew Williams Saul and Laurens Lavrysen for their comments on earlier drafts of this article. Responsibility for the views expressed remains the author's own. This article is part of a larger project on positive obligations in human rights law. Funding by the Ragnar Söderbergs Foundation is gratefully acknowledged. Footnotes 1 Costa, ‘The European Court of Human Rights: Consistency of its Case-Law and Positive Obligations’ (2008) 26 Netherlands Quarterly of Human Rights 449 at 453. 2 Stoyanova, Human Trafficking and Slavery Reconsidered. Conceptual Limits and States’ Positive Obligations in European Law (2017) at 319; McQuigg, ‘Domestic Violence as a Human Rights Issue: Rumor v. Italy’ (2016) 26 European Journal of International Law 1009. 3 Koch, Human Rights as Indivisible Rights: The Protection of Socio-economic Demands under the European Convention on Human Rights (2009); Mowbray, The Development of Positive Obligations under the European Convention on Human Rights by the European Court of Human Rights (2004) at 181. 4 Generally, authors have broken down their analysis depending on the specific subject areas covered in the case law: see Haijev, ‘The Evolution of Positive Obligations under the European Convention on Human Rights – by the European Court of Human Rights’ in Spielmann, Tsirli and Voyatzis (eds), The European Convention on Human Rights: A Living Instrument. Essays in Honour of Christos L. Rozakis (2011) 207. 5 The ECtHR itself has referred to two types of positive obligations: procedural and substantive: see Öneryildiz v Turkey Application No 48939/99, Merits and Just Satisfaction, 30 November 2004 [GC] at paras 89–96; Dröge, Positive Verpflichtungen der Staaten in der Europäischen Menschenrechtskonvention (2003); Starmer, ‘Positive Obligations under the Convention’ in Jowell and Cooper (eds), Understanding Human Rights Principles (2001) 139; Mowbray, supra n 3. 6 See O’Keeffe v Ireland Application No 35810/09, Merits and Just Satisfaction, 28 January 2014 [GC], where Judges Zupanicic, Gyulumyan, Kalaydjieva, De Gaetno and Wojtyczek in their Partly Dissenting Opinion (at para 7) observed how the scope of the positive obligations is affected by the principle of subsidiary. 7 Kratochvil, ‘The Inflation of the Margin of Appreciation by the European Court of Human Rights’ (2011) 29 Netherlands Quarterly of Human Rights 324; Lambert and Others v France Application No 46043/14, Merits and Just Satisfaction, 5 June 2015 [GC] at paras 144–148 (discussion of the margin of appreciation in the context of positive obligations). 8 Christoffersen, Fair Balance: Proportionality, Subsidiarity and Primarity in the European Convention on Human Rights (2009); Klatt and Meister, The Constitutional Structure of Proportionality (2012) at 86. 9 Plattform Arzte fur das Leben v Austria Application No 10126/82, Merits and Just Satisfaction, 21 June 1988 at para 31; Mowbray, supra n 3 at 221. 10 Hakimi, ‘State Bystander Responsibility’ (2012) 21 European Journal of International Law 341 at 349. Since the Court has justified positive obligations with the principle of effectiveness, Pieter van Dijk has observed that there are no clear-cut criteria for determining whether and when effectiveness will be achieved: see Dijk, ‘“Positive Obligations” Implied in the European Convention on Human Rights: Are the States Still the ‘Masters’ of the Convention?’ in Castermans-Holleman, van Hoof and Smith (eds), The Role of the Nation-State in the 21st Century. Human Rights, International Organizations and Foreign Policy. Essays in Honour of Peter Baehr (1998) 17 at 22. On the unpredictability of the ECtHR’s reasoning in relation to positive obligations, see also the Dissenting Opinion of Judge Kalaydjieva in Söderman v Sweden Application No 5786/08, Merits and Just Satisfaction, 12 November 2013 [GC]. See also Gerards and Brems, ‘Introduction’ in Brems and Gerards (eds), Shaping Rights in the ECHR. The Role of the European Court of Human Rights in Determining the Scope of Human Rights (2014) 1. 11 Xenos, The Positive Obligations of the State under the European Convention of Human Rights (2011) at 3; Than, ‘Positive Obligations under the European Convention on Human Rights: Towards the Human Rights of Victims and Vulnerable Witnesses’ [2003] Journal of Criminal Law 165 at 178. 12 Thielbörger, ‘Positive Obligations in the ECHR after the Stoicescu Case: A Concept in Search of Content?’ [2012] European Yearbook on Human Rights 259 at 261. 13 The Court has to navigate among different tensions and considerations: see Christoffersen, ‘Individual and Constitutional Justice: Can the Power Balance of Adjudication be Reversed?’ in Christoffersen and Madsen (eds), The European Court of Human Rights between Law and Politics (2011) 181. 14 Rights have a ‘dynamic aspect’ since they can generate new duties, which is fundamental for the understanding of their nature and function: see Raz, ‘On the Nature of Rights’ (1984) XCIII Mind 194 at 200. The strength of this approach is that it allows for flexibility for the emergence of new duties and the adaptation of the duties in light of the context and societal developments: see Report on the Right to Adequate Food as a Human Right submitted by Mr Asbjørn Eide, E/CN.4/Sub.2/1987/23, 7 July 1987 at para 47. On the dynamic aspect of rights, see also Gerards, ‘The Prism of Fundamental Rights’ (2008) 8 European Constitutional Law Review 173 at 178. 15 See also Hakimi, supra n 10 at 341, who has tried to identify such elements in the context of positive obligations under international law more generally. 16 McGrogan, ‘The Problem of Causality in International Human Rights Law’ (2016) 65 International and Comparative Law Quarterly 615 (the author argues that the UN monitoring system should focus on what can be known ‘rather than on abstract, aggregated quantitative measurements where causality cannot be plausibly attributed.’); Rigaux, ‘International Responsibility and the Principle of Causality’ in Ragazzi (ed.), International Responsibility Today. Essays in Memory of Oscar Schachter (2005) 81; Lavrysen, Human Rights in a Positive State (2016) at 137. 17 Ho and Rubin, ‘Credible Causal Interference for Empirical Legal Studies’ (2011) 7 Annual Review of Law and Social Science 17. 18 Fairgrieve, ‘Pushing the Boundaries of Public Authority Liability’ in Fairgrieve, Andenas and Bell (eds), Tort Liability of Public Authorities in Comparative Perspective (2002) 475 at 494; Booth and Squires, The Negligence Liability of Public Authorities (2006). 19 Husak, ‘Omissions, Causation and Liability’ (1980) 30 Philosophical Quarterly 318. 20 McGrath, ‘Causation by Omissions: A Dilemma’ (2005) 123 Philosophical Studies 125. 21 Steel, ‘Causation in Tort Law and Criminal Law: Unity and Divergence?’ in Dyson (ed.), Unravelling Tort and Crime (2014) 239; Fumerton and Kress, ‘Causation and the Law: Preemption, Lawful Sufficiency, and Causal Sufficiency’ (2001) 64 Law and Contemporary Problems 83. 22 Such an area is, for example, the extraterritorial application of positive obligations: see Ilascu and Others v Moldova and Russia Application No 48787/99, Merits and Just Satisfaction, 8 July 2004 [GC] at para 317. 23 L.C.B. v United Kingdom Application No 14/1997/798/1001, Merits and Just Satisfaction, 9 June 1998 at para 40. 24 Botta v Italy Application No 153/1996/772/973, Merits and Just Satisfaction, 24 February 1998 at paras 34–35. 25 Conforti, ‘Exploring the Strasbourg Case-Law: Reflections on the State Responsibility for the Breach of Positive Obligations’ in Fitzmaurice and Sarooshi (eds), Issues of State Responsibility before International Judicial Institutions (2004) 129 at 134–5 (emphasis added); Conforti, ‘Reflections on State Responsibility for the Breach of Positive Obligations: The Case-Law of the European Court of Human Rights’ (2003) 13 Italian Yearbook of International Law 3 at 3. 26 Rantsev v Cyprus and Russia Application No 25965/04, Merits and Just Satisfaction, 7 January 2010 at para 222; Mastromatteo v Italy, Application No 37703/97, Merits and Just Satisfaction, 24 October 2002 [GC] at para 76. See also Stoyanova, ‘Dancing on the Borders of Article 4: Human Trafficking and the European Court of Human Rights in the Rantsev Case’ (2012) 30 Netherlands Quarterly of Human Rights 163. 27 Xenos has also linked the elements of knowledge and causation: see Xenos, supra n 11 at 76. 28 O’Keeffe v Ireland, supra n 6 at para 144. 29 See Section 7.A where the obligation to take protective operational measures is addressed. 30 Brincat and Others v Malta Application No 60908/11, Merits and Just Satisfaction, 24 July 2014 at para 111. 31 The Court sometimes confuses the two contexts. See Section 7.B below. 32 K.U. v Finland Application No 2872/02, Merits and Just Satisfaction, 2 December 2008 at para 48. 33 In O’Keeffe v Ireland, supra n 6 at para 168, the GC established that the respondent state was aware in the 1970s of risks associated with sexual abuse of children by adults through, inter alia, ‘its prosecution of such crimes at a significant rate’. A similar approach was applied in Brincat and Others v Malta, supra n 30 at para 105, where the Court accepted that as early as 1987 laws were adopted to protect employees from asbestos and therefore since that date the state knew about the dangers associated with this substance. 34 Brincat and Others v Malta, supra n 30 at para 106. 35 Nencheva v Bulgaria Application No 48609/06, Merits and Just Satisfaction, 18 June 2013 at paras 121–2. 36 Cevriogli v Turkey Application No 69546/12, Merits and Just Satisfaction, 4 October 2016 at para 68. 37 Lavrysen, ‘Protection by the Law: The Positive Obligation to Develop a Legal Framework to Adequately Protect the ECHR Rights’ in Brems and Haeck (eds), Human Rights and Civil Rights in the 21st Century (2014) 69. 38 In some circumstances an initial assumption that the authorities had duties, and therefore ought to have known about risks of harm, might be warranted: see Premininy v Russia Application No 44973/04, 10 February 2011 at para 85 (prisoner beaten by other prisoners). 39 Van Colle v United Kingdom Application No 7678/09, Merits and Just Satisfaction, 13 November 2012 at para 96, where the Court determined that the harm was not foreseeable in the first place. See also Hiller v Austria, Application No 1967/14, Merits and Just Satisfaction, 22 November 2016 at para 53 (the psychiatric hospital staff could not have foreseen that a patient will escape and commit suicide). 40 Ugrekhelidze, ‘Causation: Reflection in the Mirror of the European Convention on Human Rights (A Sketch)’ in Calflisch, Callewaert, Liddell, Mahoney and Villeger (eds), Liber Amicorum Luzius Wildhaber Human Rights–Strasbourg Views (2007) 469 at 476. 41 See E. and Others v United Kingdom Application No 33218/96, Merits and Just Satisfaction, 26 November 2002 at para 99. 42 Salakhov and Islyamova v Ukraine Application No 28005/08, Merits and Just Satisfaction, 14 March 2013 at para 181 (‘Whether or not the authorities’ efforts could in principle have averted the fatal outcome in the present case is not decisive for this conclusion [failure to discharge a positive obligation]. What matters for the Court is whether they did everything reasonably possible in the circumstances, in good faith and in a timely manner, to try to save the first applicant’s life.’) 43 O’Keeffe v Ireland, supra n 6 at para 149; Opuz v Turkey Application No 33401/02, Merits and Just Satisfaction, 9 June 2009 at para 136; Premininy v Russia Application No 44973/04, Merits and Just Satisfaction, 10 February 2011 at para 84; Bljakaj and Others v Croatia Application No 74448/12, Merits and Just Satisfaction, 18 September 2014 at para 124 [emphasis added]. 44 Supra n 6 at para 166. See also E. and Others v United Kingdom supra n 41 at para 100. 45 Application No 52806/09 and 22703/10, Merits and Just Satisfaction, 5 December 2013 at paras 225 and 229. 46 Applications Nos 15339/02 et al., Merits and Just Satisfaction, 20 March 2008 at para 158. 47 Dodov v Bulgaria Application No 59548/00, Merits and Just Satisfaction, 17 January 2008 at para 97. 48 Draon v France Application No 1513/03, Merits and Just Satisfaction, 6 October 2005 [GC] at para 106. A clarification as to the application of the ‘direct and immediate link’ test in the context of Article 8 is due here. The Court seems to use the test to also determine whether the definitional threshold of Article 8 can be engaged in the first place. This can be related to the tendency of collapsing the definitional threshold enquiry, on the one hand, with the enquiry about the triggering and scope of positive obligations, on the other: see Gerards and Senden, ‘The Structure of Fundamental Rights and the European Court of Human Rights’ (2009) 7 International Journal of Constitutional Law 619. 49 Dubetska and Others v Ukraine Application No 30499/03, Merits and Just Satisfaction, 10 February 2011 at para 123. 50 Fadeyeva v Russia Application No 55723/00, Merits and Just Satisfaction, 9 June 2005 at para 92 [emphasis added]. 51 Stoyanovi v Bulgaria Application No 42980/04, Merits and Just Satisfaction, 9 November 2010 at para 61. 52 Giuliani and Gaggio v Italy Application No 23458/02, Merits and Just Satisfaction, 24 March 2011 [GC] at para 253. 53 Supra n 41 at para 100. 54 Application No 50385/99, Merits and Just Satisfaction, 20 December 2004 [GC] at para 63 [emphasis added]. 55 Application No 1967/14, Merits and Just Satisfaction, 22 November 2016 at para 53. 56 Another way in which causation matters in human rights law and in international law more generally, relates to remedies. In particular, once a violation of a right has been found, causation needs to be established between the violation and any harm for the purposes of awarding damages. See Article 41 ECHR. For an in-depth discussion on causation in the context of Article 41, see Kellner and Durant, ‘Causation’ in Fenyves, Karner, Koziol and Steinder (eds), Tort Law in the Jurisprudence of the European Court of Human Rights (2011) 449. This is an area of enquiry not pursued in this article. 57 Crawford, State Responsibility. The General Part (2013) at 113. 58 Draft Articles on Responsibility of States for International Wrongful Acts with Commentaries, Yearbook of International Law Commission, 2001, Vol II (Part Two) at 38, para 2. 59 Articles 4–7 ILC Draft Articles form the hard core of the doctrine of attribution since they deal with organs and agencies of state exercising sovereignty authority: see Crawford, supra n 57 at 115. 60 See Article 4(2) ILC Draft Articles; and Crawford, ibid. at 126. 61 Article 8 ILC Draft Articles. 62 See Article 7 ILC Draft Articles; and Armed Activities (DRC v Uganda) ICJ Reports 2005, 162 at 242. The difficulty here lay in distinguishing an official, though ultra vires, act from a purely private act: see Crawford, supra n 57 at 115. 63 Article 7 Draft Articles Commentary at para 18; and ILC Commentary, supra n 58 at 43, para 7. 64 I say ‘might’ because ‘[a]s a normative operation, attribution must be clearly distinguished from the characterization of conduct as internationally wrongful’: see ILC Commentary, supra n 58 at 39, para 4. 65 ILC Commentary, supra n 58 at 39 para 4. 66 ILC Commentary, supra n 58 at 47 para 1. 67 Contrast Military and Paramilitary Activities in against Nicaragua (Nicaragua v US) (Merits) [1986] ICJ Reports 14 at paras 109–115 (requiring effective control) with Case IT-94-1, Prosecutor v Tadic  (Appeal Judgement) IT-94-1-A, International Criminal Tribunal for the former Yugoslavia (ICTY), 15 July 1999 at paras 115-131 (adopting a standard of overall control). See also the confirmation of the effective control test in Application of the Convention on the Prevention and Punishment of the Crime of Genocide, ICJ Reports 2007, 43 at 207; and DRC v Uganda, supra n 62 at 226. 68 International law has accepted the distinction between primary and secondary rules of state responsibility. Primary rules are the substantive obligations in the various subject areas of international law. Secondary rules are those that elaborate on what it means for a state to be held responsible for violations of these duties. The rules on attribution belong to the latter. Positive obligations belong to the former: see Draft Articles on State Responsibility, supra n 58 at para 1. 69 It needs to be acknowledged, however, that on some occasions the ECtHR is not clear in its judgments as to why the respondent state can be held responsible: is it because harmful conduct is attributable to it or is it because it failed to fulfill its positive obligations: see Cyprus v Turkey Application No 25781/94, 10 May 2011 at paras 69–80. For discussion on the Court’s unclear logic, see Milanovic, ‘From Compromise to Principle: Clarifying the Concept of State “Jurisdiction” in Human Rights Treaties’ (2008) 8 Human Rights Law Review 411. 70 Ilascu and Others v Moldova and Russia supra n 22 at para 319. 71 Report of the ILC A/56/10 (2001) ‘Draft Articles’ at 44, (‘State’s authorities are strictly liable for the conduct of their subordinates; they are under a duty to impose their will and cannot shelter behind their inability to ensure that it is respected’). See also ibid. at para 319. 72 Report of the ILC A/56/10 (2001) Draft Articles at 102. 73 Evans, ‘State Responsibility and the European Convention on Human Rights: Role and Realm’ in Fitzmaurice and Sarooshi (eds), Issues of State Responsibility before International Judicial Institutions (2004) 139 at 157. 74 Medova v Russia Application No 25385/04, Merits and Just Satisfaction, 15 January 2009 at para 95 (disappearance case); Albekov and Others v Russia Application No 68216/01, Merits and Just Satisfaction, 9 October 2008 at paras 80–86 (it was not possible to establish who laid the mines which caused death; the Court did not have to decide on the issue since the respondent state was aware that mines were laid in the area and were under the positive obligation to protect the residents from the risks involved). 75 In the context of human rights law, there is a jurisdiction threshold which has to be passed so that the application of human rights treaties is triggered in the first place: see Milanovic, Extraterritorial Application of Human Rigts Treaties: Law, Principles, and Policy (2011). Control might be crucial for passing this threshold: see Lawson, ‘Life After Bankovic: On the Extraterritorial Application of the European Convention on Human Rights’ in Coomans and Kamminga (eds), Extraterritorial Application of Human Rights Treaties (2004) 83. 76 For this distinction, see Lawson, ‘Out of Control. State Responsibility and Human Rights: Will the ILC’s Definition of the “Act of State” Meet the Challenges of the 21st Century’ in Castermans-Holleman, Hoof and Smith (eds), supra n 10, 91 at 97. 77 Similar considerations have been made relevant in the context of claims against public authorities for damages at national level: see Brodie, ‘Compulsory Altruism and Public Authorities’ in Fairgrieve, Andenas and Bell (eds), Tort Liability of Public Authorities in Comparative Perspective (2002) 541 at 551 (‘I would suggest that where a public authority is concerned the court should, rather than seeking to identify pure omissions, look to see whether the authority in question has statutory responsibility to control, regulate, or supervise the relevant area of social or economic activity in the community’). 78 Sargyan v Azerbaijan Application No 40167/06, Merits, 16 June 2015 [GC] at paras 128–131; Besson, ‘The Bearers of Human Rights; Duties and Responsibilities for Human Rights: A Quite (R)evolution’ (2015) 32 Social Philosophy and Policy 244 at 253 (‘there is a general human rights’ positive duty for states to exercise jurisdiction and hence to incur human rights duties’). 79 Sargyan v Azerbaijan, ibid. at para 131. The following clarification is due here. In Sargyan v Azerbaijan, the issue under discussion was control for the purpose of establishing jurisdiction under Article 1 ECHR. However, the same logic can be extended in the context of positive obligations. On the use of the notion of ‘control’ for different purposes (establishing jurisdiction, attribution and positive obligations), see Besson, ‘Concurrent Responsibilities under the European Convention on Human Rights: The Concurrence of Human Rights Jurisdictions, Duties and Responsibilities’ in van Aaken and Motoc (eds), The European Convention on Human Rights and General International Law (forthcoming). 80 Finogenov and Others v Russia Application 18299/03 and 27311/03, 20 December 2011 at para 208. 81 See Article 2(2) of the ECHR which delineates limited circumstances when use of force might be justified. 82 McCann and Others v United Kingdom Application No 18984/91, Merits and Just Satisfaction, 27 September 1995 [GC] at para 194. 83 Supra n 52 at paras 253–262. 84 Application No 57950/00, Merits and Just Satisfaction, 24 February 2005 at para 188. 85 Supra n 54 at para 69. 86 Ibid. at para 70. 87 Supra n 80 at 209. 88 Ibid. at para 213 [emphasis added]. 89 Ibid. at para 214. 90 Ibid. at para 243. See also Mikayil Mammadov v Azerbaijan Application No 4762/05, Merits and Just Satisfaction, 17 December 2009 at paras 116–119, where the applicant complained that police officers were responsible for his wife’s death because upon their entry into the family’s dwelling she set herself on fire. The degree to which the state agents had control over the situation and the related degree to which they could have predicted the course of the events were essential in determining the scope of positive obligations under Article 2. 91 This paradox is further exaggerated if one takes into account that human rights are traditionally invoked to circumscribe the exercise of state control (states’ negative obligations to refrain), while positive human rights obligations might require more state control: see Reus-Smit, ‘On Rights and Institutions’ in Beits and Goodin (eds), Global Basic Rights (2011) 26 at 44. 92 Costello-Roberts v United Kingdom Application No 13134/87, Merits and Just Satisfaction, 25 March 1993; Dodov v Bulgaria, supra n 47; Storck v Germany Application No 61603/00, Merits and Just Satisfaction, 16 June 2005 at para 103; O’Keeffe v Ireland, supra n 6 at para 150. 93 Calvelli and Ciglio v Italy Application No 32967/96, Merits, 17 January 2002 [GC] at para 49; Vo v France Application No 53924/00, Merits, 8 July 2004 [GC] at para 89; Center for Legal Resources on behalf of Valentin Câmpeany v Romania Application No 47848/08, Merits and Just Satisfaction, 17 July 2014 [GC] at para 130; Lambert and Others v France Application No 46043/14, Merits and Just Satisfaction, 5 June 2015 [GC] at para 140 [emphasis added]. 94 One way will be to say that public functions are the ones that the state has historically performed. For a more in depth discussion, see Thomas, Public Rights, Private Relations (2015) at 41. 95 Supra n 6 at para 151. 96 Ibid. at para 162. This notion of ‘inherent obligation’ has not remained unchallenged. In their dissenting opinion, five judges challenged the assumption that there is some inherent risk of sexual abuse in the context of education (at para 15). 97 Ibid. at para 145 [emphasis added]. 98 Ibid. at para 157. 99 Application No 19986/06, Merits and Just Satisfaction, 10 April 2010 at para 35. 100 Storck v Germany, supra n 92 at para 103. 101 Dodov v Bulgaria Application No 59548/00, Merits and Just Satisfaction, 17 January 2008 at paras 84–86. 102 Application No 19202/03, Merits and Just Satisfaction, 24 April 2012 at para 63 (basic safety precautions were missing including not locking the door of the transformer located at a park for children). 103 Supra n 36 at para 57. 104 Ibid. at para 67. 105 Ibid. at para 69. 106 Reference was made to the various texts adopted by the Council of Europe in the field of environment and the industrial activities: Öneryildiz v Turkey, supra n 5 at paras 59–62 and 71. 107 Ibid. at para 71. Other industrial activities reviewed by the Court are nuclear testing (L.C.B. v United Kingdom Application No 14/1997/798/1001, Merits and Just Satisfaction, 9 June 1998), toxic emission from fertilizer factory (Guerra and Others v Italy Application No 14967/89, Merits and Just Satisfaction, 19 February 1998 [GC]) and exposure to asbestos at a workplace (Brincat and Others v Malta, supra n 30). 108 Vilnes and Others v Norway, supra n 45 at para 223. 109 Kolyadenko and Others v Russia Application No 17423/05, Merits and Just Satisfaction, 28 February 2012 at para 164. 110 Brincat and Others v Malta, supra n 30 at para 81 [emphasis added]. 111 Vilnes and Others v Norway, supra n 45 at para 223 (the Court ‘sees no need to consider in detail the degree of involvement of the respondent State in the hazardous activity in question, since the Convention obligation applies to “any activity, whether public or private”’). 112 Budayeva and Others v Russia, supra n 46 at para 174. The same distinction between man-made harm and natural harm is very clearly reflected in the Court's non-refoulement case law under Article 3 of the ECHR: see Stoyanova, ‘How Exceptional Must “Very Exceptional” Be? Non-Refoulement, Socio-Economic Deprivation and Paposhvili v Belgium’ (2017) 29 International Journal of Refugee Law 580. 113 Ibid. at para 137. 114 Ibid. at para 135 [emphasis added]. 115 Ibid. at paras 158-159. 116 Supra n 5 at para 134 [emphasis added]. 117 Ibid. at para 135. 118 Budayeva and Others v Russia, supra n 46 at para 173. 119 Ibid. at para 174. 120 Ibid. at paras 76–77. 121 I use the term ‘victim’ here in a general sense as a person who has sustained harm without prejudice to the determination whether the state can be held responsible under ECHR for this harm. 122 Keenan v United Kingdom Application No 27229/95, Merits and Just Satisfaction, 3 April 2001 at para 91. 123 Paul and Audrey Edwards v United Kingdom Application No 46477/99, Merits and Just Satisfaction, 14 March 2012 at paras 57–64 (a detainee was killed by another detainee while held in prison); Premininy v Russia Application No 44973/04, Merits and Just Satisfaction, 10 February 2011 at para 91 (a detainee was systematically beaten by other detainees). 124 Keenan v United Kingdom, supra n 122 at para 90 (the applicant argued that her son died from suicide due to the prison authorities failure to protect his life). 125 Florea v Romania Application No 37186/03, Merits and Just Satisfaction, 14 September 2010 (protection from passive smoking); Jashi v Georgia Application No 10799/06, Merits and Just Satisfaction, 8 January 2013 (provision of adequate care for detainee’s mental health). 126 Oette, ‘Austerity and the Limits of Policy-Induced Suffering: What Role for the Prohibition of Torture and Other Ill-Treatment?’ (2015) 15 Human Rights Law Review 669 at 681; O’Cinneide, ‘A Modest Proposal: Destitution, State Responsibility and the European Convention on Human Rights’ (2008) 5 European Human Rights Law Review 583. 127 Waiter v Poland Application No 42290/08, Admissibility, 15 May 2012 at paras 36-42 (access to life-saving drug); Budina v Russia Application No 45603/05, Admissibility, 18 June 2009. 128 Gerards, ‘The ECtHR’s Response to Fundamental Rights Issues related to Financial and Economic Difficulties: The Problem of Compartmentalization’ (2015) 13 Netherlands Quarterly of Human Rights 274. 129 Orchowski v Poland Application No 17885/04, Merits and Just Satisfaction, 22 October 2009 at para 153. 130 See also Nencheva v Bulgaria Application No 48609/06, Merits and Just Satisfaction, 18 June 2013 at para 119 (severely disabled children held in an institution died during the winter); Centre for Legal Resources on Behalf of Velentin Campeanu v Romania, supra n 93 at paras 134–144. This might also explain the Court’s lenient approach to causation in relation to victims of human trafficking held in immigration detention in the context of the application of Article 4 ECHR: see Stoyanova, ‘L.E. v. Greece: Human Trafficking and the Scope of States’ Positive Obligations under the ECHR’ (2016) 3 European Human Rights Law Review 290 at 299. 131 Denis Vasilyev v Russia Application No 32704/04, Merits and Just Satisfaction, 17 December 2009 at paras 115–116. 132 See Adam, R v Secretary of the State for the Home Department [2005] UKHL 66 (depriving asylum-seekers from social support when they are not allowed to engage in remunerated employment). 133 Keller v Russia Application No 26824/04, Merits and Just Satisfaction, 17 October 2013 at para 88. 134 Nencheva and Others v Bulgaria Application No 48609/06, Merits and Just Satisfaction, 18 June 2013 at paras 121–122. 135 Peroni and Timmer, ‘Vulnerable Groups: The Promise of an Emerging Concept in European Human Rights Convention Law’ (2013) 11 International Journal of Constitutional Law 1056; Ippolito and Sanchez (eds), Protecting Vulnerable Groups: The European Human Rights Framework (2015). 136 E. and Others v United Kingdom, supra n 41 at para 88. 137 M.S.S. v Belgium and Greece Application No 30696/09, Merits and Just Satisfaction, 21 January 2011 [GC] at para 232. 138 Placi v Italy Application No 48754/11, Merits and Just Satisfaction, 21 January 2014 at para 49. 139 Storck v Germany, supra n 92 at para 105. 140 Bevacqua and S. v Bulgaria Application No 71127/01, Merits and Just Satisfaction, 12 June 2008 at para 65. 141 D.H. and Others v Czech Republic Application No 57325/00, Merits and Just Satisfaction, 13 November 2007 [GC] at para 182. 142 Milanovic v Serbia Application No 44614/07, Merits and Just Satisfaction, 14 December 2010 at para 89. 143 One can distinguish this situation from ‘pre-existing vulnerabilities.’ For a discussion in the context of tort law, see Green, Causation in Negligence (2015) at 38. 144 I v Finland Application No 20511/03, Merits and Just Satisfaction, 17 July 2008 at para 44. 145 Cyprus v Turkey Application No 25781/94, Merits, 10 May 2001 [GC] at para 219; Waiter v Poland, supra n 127 at para 35. 146 Lopes de Sousa Fernandes v Portugal Application No 56080/13, Merits and Just Satisfaction, 15 December 2015 at para 114; Elena Cojocaru v Romania Application No 74114/12, Merits and Just Satisfaction, 22 March 2016 at para 111. 147 Ibid. at para 188. 148 Application No 55164/08, Merits and Just Satisfaction, 14 October 2010 at para 79. 149 Application No 46117/99, Merits and Just Satisfaction, 10 November 2004 (the national authorities did not comply with decisions by the national court ordering the closure of a mine); Giacomelli v Italy Application No 59909/00, Merits and Just Satisfaction, 2 November 2006 at para 93 (the state authorities did not comply with domestic legislation on environmental matters and failed to enforce judicial decisions); Kalender v Turkey Application No 4314/02, Merits and Just Satisfaction, 15 December 2009 at paras 43–47 (relatives of the applicants were killed in an accident at a railway station; the Court found a violation of Article 2 in its substantive aspect in view of the significant number and the seriousness of the breaches of the national safety regulations). 150 See generally, Lavrysen, supra n 37 at 82. 151 See, for example, Panaitescu v Romania Application No 30909/06, Merits and Just Satisfaction, 10 April 2012 at para 36. 152 Fadeyeva v Russia, supra n 50 at para 98 (after observing that the domestic legality is not a separate and conclusive test, the Court added that it is ‘rather one of many aspects which should be taken into account in assessing whether the State has struck a “fair balance” in accordance with Article 8(2)’). The problem with this approach is that the Court combines various factors, including non-compliance with the national legislation, into the general and very elusive standard of fair balance. This makes it impossible to objectively assess the role of and the weight attached to each individual factor, including non-compliance with domestic legislation. 153 Gerards and Senden, supra n 48 at 619. 154 Brems and Lavrysen, ‘Procedural Justice in Human Rights Adjudication: the European Court of Human Rights’ (2013) 35 Human Rights Quarterly 182; Brems, ‘Procedural Protection. An Examination of Procedural Safeguards Read into Substantive Convention Rights’ in Brems and Gerards (eds), Shaping Rights in the ECHR. The Role of the European Court of Human Rights in Determining the Scope of Human Rights (2014) 137; Gerards and Brems (eds), Procedural Review in European Fundamental Rights Cases (2017). 155 Öneryildiz v Turkey, supra n 5 at paras 89–90; Budayeva and Others v Russia, supra n 46 at para 132; Roche v United Kingdom Application No 32555/96, Merits and Just Satisfaction, 19 October 2005 [GC] at paras 161–162; Hatton and Others v United Kingdom Application No 36022/97, Merits and Just Satisfaction, 8 July 1997 [GC] at para 104. See also Steyn and Slarks, ‘Positive Obligation to Provide Access to Information under the European Convention on Human Rights’ (2012) 17 Judicial Review 308. 156 Roche v United Kingdom, ibid. at para 162. 157 Ibid. at para 161. 158 Hatton and Others v United Kingdom, supra n 155. 159 Ibid. at para 128. 160 Vilnes and Others v Norway, supra n 45 at paras 233–244. 161 Diving tables relate to the planning and the monitoring of the decompression. 162 Vilnes and Others v Norway, supra n 45 at para 233 [emphasis added]. 163 Ibid. at para 240. 164 Ibid. at para 238. 165 Ibid. at para 243. 166 Ibid. at para 244 [emphasis added]. 167 O’Keeffe v Ireland, supra n 6 at para 144. 168 See Hickman, ‘The Reasonableness Principle: Reassessing its Place in the Public Sphere’ (2004) 63 Cambridge Law Journal 166. 169 Mastromatteo v Italy, supra n 26 at para 74. 170 Ibid. at para 76 [emphasis added]. 171 For a useful outline of the interests involved, see the Partly Dissenting Opinion of Judge Bonello in Mastromatteo v Italy, supra n 26 at para 7. 172 Osman v United Kingdom Application No 23452/94, Merits and Just Satisfaction, 28 October 1998 [GC] at para 116. 173 Gerry, ‘Obligation to Prevent Crime and to Protect and Provide Redress to Victims of Crime’ in Colvin and Cooper (eds), Human Rights in the Investigation and Prosecution of Crime (2009) 423 at 432. 174 The interpretation advanced by McBride that the likelihood ‘has at least to be compelling’ is too stringent: see McBride, ‘Protecting Life: Positive Obligation to Help’ (1999) 24 European Law Review 43. 175 See Re W’s Application [2004] NIQB 67; Re Officer L [2007] UKHL 36 per Lord Carswell; Smith v Chief Constable of Sussex [2008] EWCA Civ 39; and Wright, ‘The Operational obligation under Article 2 of the European Convention on Human Rights and Challenges of Coherence–Views from English Supreme Court and Strasbourg’ (2016) 1 Journal of European Tort Law 58 at 67. 176 Ebert and Sijniensky, ‘Preventing Violation of the Right to Life in the European and the Inter-American Human Rights Systems: From Osman Test to a Coherent Doctrine on Risk Prevention?’ (2015) 15 Human Rights Law Review 343 at 359. 177 McBride, supra n 174. Xenos has also objected to the general applicability of the ‘real and immediate’ test. However, the substantiation of his objection is questionable. More specifically, he has observed that ‘the Osman test contains calculating parameters of “seriousness” and “immediacy” that confine the scope of positive obligations to mere reactive responses’ [emphasis added]. National authorities are, however, under the obligation to take protective operational measures upon risk of harm and before the harm has actually materialized. Certainly, once a case reaches the ECtHR, then an assessment of the national authorities’ reaction is made: see Xenos, supra n 11 at 112. 178 Opuz v Turkey, supra n 43 at paras 133-136; Rantsev v Cyprus and Russia, supra n 26 at para 286; Branko Tomasic and Others v Croatia Application No 46598/06, Merits and Just Satisfaction, 15 January 2009 at para 52. 179 See the Partly Dissenting Opinion of Judge Bonello in Mastromatteo v Italy, supra n 26. 180 In some judgments, the Court incorrectly applies the ‘real and immediate risk’ test. For example, in Öneryilidz v Turkey, supra n 5 at paras 100–102, the Court applied this test in its assessment as to whether Turkey was responsible for the deaths in the light of the substantive aspect of Article 2, that is, whether the national safety regulations were effective. The reference to the test in this context was, however, hard to defend. While the establishment that the national authorities knew about the dangers associated with the garbage collection point was necessary for the finding that state omissions were causative to harm (see Section 3 above), it is hard to understand why it needed to be also established that these dangers were immediate in the sense of the Osman test. In Öneryilidz v Turkey the notion of immediacy seems to be stretched to a breaking point (at para 100): ‘the risk of an explosion had clearly come into being long before it was highlighted in the report of 7 May 1991 and that, as the site continued to operate in the same conditions, that risk could only have increased during the period until it materialized on 28 April 1993.’ A period of two years is hardly what the Court had in mind when it established the ‘real and immediate risk’ test in Osman v United Kingdom. 181 There are judgments where the Court is very clear about the distinction between circumstances calling for protective operational measures and circumstances requiring general protection of the society: see Bljakaj and Others v Croatia Application No 74448/12, Merits and Just Satisfaction, 18 September 2014 at para 124; Stoyanovi v Bulgaria Application No 42980/04, Merits and Just Satisfaction, 9 November 2010 at paras 59 and 62; Mikhno v Ukrain Application No 32514/12, Merits and Just Satisfaction, 1 September 2016 at para 126. 182 For a useful outline, see Lavrysen, supra n 37. 183 See the Partly Concurring and Partly Dissenting Opinion of Judge Sajó in Banel v Lithuania Application No 14326/11, Merits and Just Satisfaction, 18 June 2013 (death of a boy after collapse of a roof). 184 The issue of acceptable level of risk has come to a head in the context of home births and state imposed regulations as to the conditions under which women can give birth: see the Dissenting Opinion of Judges Sajó, Karakas, Nicolaou, Laffranque and Kellder in Dubská and Krejzová v Czech Republic Applications Nos 28859/11 and 28473/12, Merits and Just Satisfaction, 15 November 2016 [GC] at para 29. 185 See Mastromatteo v Italy, supra n 26 at para 73, where it was observed that the percentage of crimes committed by prisoners on semi-custodial regime was very low. 186 Lopes de Sousa Fernandes v Portugal, supra n 146 at para 108; Lopes de Fernandes v Portugal, supra n 147 at paras 168 and 187; Elena Cojocaru v Romania, supra n 146 at para 100. Similar reasoning has been endorsed in other contexts too: see Stoyanovi v Bulgaria. supra n 181 at para 61. 187 As opposed to the Chamber in Lopes de Sousa Fernandes, supra n 146, which found a substantive violation of Article 2, the Grand Chamber, supra n 147, did not, in this way overruling the Chamber. For a more detailed analysis, see Lavrysen, ‘Medical Negligence after Lopes de Sousa Fernandes: A Blank Check to the Member States with Respect to the Substantive Right to Life?’ Strasbourg Observers Blog, 8 February 2018. 188 Lopes de Sousa Fernandes v Portugal, supra n 147 at paras 182, 190–192. In paras 191 and 192 of the judgment, the Grand Chamber explained the circumstances when ‘denial of immediate emergency care’ will transpire. These circumstances are framed as ‘very exceptional circumstances’. On the application of the ‘very exceptional circumstances’ standards, see Stoyanova, ‘How Exceptional Must “Very Exceptional” be? Non-Refoulement, Socio-Economic Deprivation and Paposhvili v Belgium’ (2017) 29 International Journal of Refugee Law 580. 189 Supra n 10 at para 50. 190 Ibid. at para 91 [emphasis added]. 191 The standards have not been applied in other cases either since Söderman v Sweden, ibid. 192 Ibid. at para 117. 193 O’Keeffe v Ireland, supra n 6 at para 166. 194 In Söderman v Sweden, supra n 10, the applicant invoked the ineffectiveness of not only the national criminal law remedies, but also of the civil law remedies. However, the civil law remedies were contingent on the criminal law remedies. 195 M.C. v Bulgaria Application No 39272/98, Merits and Just Satisfaction, 4 December 2003; Siliadin v France Application No 73316/01, Merits and Just Satisfaction, 26 July 2005. See generally Stoyanova, supra n 2 at 330; Lazarus, ‘Positive Obligations and Criminal Justice: Duties to Protect or Coerce’ in Zadner and Roberts (eds), Principles and Values in Criminal Law and Criminal Justice: Essays in Honour of Andrew Ashworth (2012) 136 at 150; Stoyanova, ‘Article 4 of the ECHR and the Obligation of Criminalising Slavery, Servitude, Forced Labour and Human Trafficking’ (2014) 3 Cambridge Journal of International and Comparative Law 407 at 414. 196 In the circumstances of unintentional killing, the Court has clarified that absence of a criminal law remedy might not be problematic: see Calvelli and Ciglio v Italy, supra n 93; and Vo v France, supra n 93. 197 Söderman v Sweden, supra n 10 at para 82. 198 Belenko v Russia Application No 25435/06, Merits and Just Satisfaction, 18 December 2014 at paras 84–85; Burzykowski v Poland Application No 11562/05, Merits and Just Satisfaction, 27 June 2006 at para 118; Dodov v Bulgaria, supra n 47 at para 98; Kudra v Croatia Application No 13904/07, Merits and Just Satisfaction, 18 December 2012 at paras 106–121; Bilbija and Blazevic v Croatia Application No 62870/13, Merits and Just Satisfaction, 12 January 2016 at para 119. © The Author(s) [2018]. Published by Oxford University Press. All rights reserved. For permissions, please email: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Human Rights Law Review Oxford University Press

Causation between State Omission and Harm within the Framework of Positive Obligations under the European Convention on Human Rights

Human Rights Law Review , Volume Advance Article (2) – May 25, 2018

Loading next page...
 
/lp/ou_press/causation-between-state-omission-and-harm-within-the-framework-of-CjLoE0Icqt
Publisher
Oxford University Press
Copyright
© The Author(s) [2018]. Published by Oxford University Press. All rights reserved. For permissions, please email: journals.permissions@oup.com
ISSN
1461-7781
eISSN
1744-1021
D.O.I.
10.1093/hrlr/ngy004
Publisher site
See Article on Publisher Site

Abstract

Abstract The issue of causation has been surprisingly overlooked in the area of international human rights law. The objective of this article is to fill this gap by investigating how the ECtHR finds causal connections between harm and state omissions within the framework of positive obligations. By engaging with causation, this article seeks to partially address the widely voiced concerns about the indeterminacy that clouds positive obligations in the case law. Four main arguments are articulated. First, assessments whether the state knew, or ought to have known, about the (risk of) harm, whether demanding state action is reasonable and whether harm is caused by state failures, are merged and affect each other in the enquiry as to whether the state has failed to fulfill its positive obligations. Secondly, the level of state control structures lines of causation. Thirdly, since the question as to how much control the state should have could imply normative judgments in which the Court might not want to see itself implicated, and since empirical and epistemological uncertainly might hamper assessments of causation, the Court has recourse to techniques to avoid direct resolution of these normative issues and uncertainties. Two such techniques are discussed: domestic legality and national procedural guarantees. Finally, even in cases where omissions might be causative to harm, additional considerations might militate against finding the state responsible under the ECHR: reasonableness, no immediacy of the harm and no systemic failures. 1. INTRODUCTION Positive obligations have penetrated all parts of the European Convention on Human Rights (ECHR or the Convention) and there are no a priori limits to the contexts in which they may be found to arise.1 Due to their considerable importance, scholarly contributions have focused on different areas covered by positive obligations, such as domestic violence and other forms of interpersonal violence,2 environmental pollution3 or socio-economic assistance.4 Efforts to systematize positive obligations as developed in the case law have also been made.5 There has also been extensive discussion of certain principles that affect these obligations, such as subsidiarity,6 margin of appreciation7 and proportionality.8 The role of causation, however, has been hitherto largely neglected. No analysis has been offered as to how the Court (ECtHR) finds causal connections between harm and state omissions. Filling this gap is the aim of this article. The importance of addressing this blind spot relates to the general uncertainty that clouds positive obligations. The ECtHR has not proposed a general analytical framework for reviewing them and has explicitly refused ‘to develop a general theory of the positive obligations which may flow from the Convention.’ 9 Perhaps as a consequence, it has been observed that the ECtHR’s approach to positive obligations is incoherent and even arbitrary, which is not conductive to certainty and predictability.10 One is left with the impression that the Court simply makes in casu judgments when dealing with positive obligations, and that it is hard to direct and structure these by extracting distinctions as to the steps taken and the principles applied. The quality of the Court’s reasoning has also been criticized. It has been noted that the problem with positive obligations is that ‘their proper scope appears open-ended’ and the Strasbourg court ‘does not set general conceptual limitations’ for its interventions in developing them.11 It has been added that such an approach ‘put[s] the concept of positive obligations into disrepute.’12 By engaging with causation, this article seeks to partially address the above concerns. While the problem of indeterminacy cannot be eliminated13 and in fact, should not be conclusively resolved,14 there are certain elements identifiable in the judgments which determine the intensity of the positive obligations.15 In this contribution, I focus on the element of causation as also related to the elements of knowledge and reasonableness. Causation implies some nexus/proximity between the harm sustained by the applicant (harm that falls within the definitional scope of one of the protected rights) and the alleged omission by the state to ensure the right. The issue of causation has been extensively addressed in other areas of law, but surprisingly neglected in the area of international human rights law.16 However, causation is essential for understanding positive obligations and responding to the above mentioned concerns as to the elusiveness of their scope. Any engagement with causation has to start with the awareness that ascribing causality in human society is fraught with complexities.17 There has been a clear acknowledgment that causality by omission is hard to ascertain. National law has struggled with issues of causality by omission18 and philosophy has also struggled with this issue.19 A further problem is that causality by omission implies a counter-factual and speculative analysis. It might be possible to identify different omissions that might have causal connections to the harm.20 The role of normativity and policy considerations when determining causality has been also noted.21 The issue of causality is also fraught with difficulties from an evidential point of view since determining causality might be a highly factual process. The ECtHR is certainly confronted with all these challenges that can be also related to the uncertainties about the scope and the intensity of the positive obligations generated by the ECHR. In this article, I will show how the ECtHR approaches these challenges. Section 2 will elaborate on the question as to why causation is significant in the context of positive obligations under the ECHR. Since the establishment of causation is related to the determination whether the state knew or should have known about harm or risk of harm, Section 3 will discuss the interrelationship between causation on the one hand, and knowledge and foreseeability on the other. Even if harm or risk of harm is foreseeable, the Court can still verify the place of state omissions in the chain of events. For this purpose, however, no clear test of causation has been articulated, as Section 4 will show. Against this backdrop, I suggest that inspiration from other areas of the law on state responsibility could be useful. Section 5 thus draws a parallel with the rules on attribution in international law. Since these rules express lines of proximity, it is useful to assess their underlying justifications so that we better understand the linkages between harm and state conduct. More specifically, the rules on attribution are founded on the principle that control implies responsibility and the same principle can be extended in the context of positive obligations. Accordingly, the degree of control exercised by the state is essential for assessing lines of causation. This is also reflected in the case law of the Court. Since the question as to how much control the state should have could imply normative judgments in which the Court might not want to see itself implicated, and since empirical and epistemological uncertainly might hamper assessments of causations, the Court can resort to techniques that avoid the direct resolution of these normative issues and uncertainties. Section 6 will identify two such techniques: domestic legality and procedural guarantees. Finally, Section 7 will demonstrate that an additional corrective can be used, namely the test of reasonableness, which permeates the whole case law on positive obligations. The last implies that even in cases of clear causal connections between harm and state omissions, it might be unreasonable to expand the scope of the positive obligations to such as an extent as to lead to state responsibility. Section 7 will also discuss two other techniques for limiting state responsibility: the ‘real and immediate risk’ test and the requirement for systemic as opposed to incidental failures. A brief note regarding methodology is due. In the selection of case law, priority is given to judgments under Articles 2, 3 and 8 of the ECHR delivered by the Grand Chamber. I have also analysed Chamber judgments, including those that are given prominence in the existing literature as heralding important developments concerning positive obligations. Although there are areas that have remained uncharted in this article,22 the selected case law can adequately serve the purpose of exploring the role of causation in the context of positive obligations. The method employed here is to both reflect upon and explain the approach taken by the Court. At certain points, however, I do take a critical approach to the case law. I extend the limits of my descriptive approach in the face of some inconsistencies within the case law and inadequacies as to how the Court has grappled with some issues that demand more serious consideration in the future. 2. WHY DOES CAUSATION MATTER? Identifying the causation between harm and state omission is crucial for finding the respondent state responsible for that omission under the ECHR. This was clearly exemplified in L.C.B. v United Kingdom, a case demonstrative of a failure to establish this connection. The applicant sought to attribute her leukemia to her father’s exposure to radiation from atmospheric tests of nuclear weapons during his military service. She claimed that the failure by the state to warn her parents of the possible risk to her health caused by her father’s participation in the nuclear tests, and its earlier failure to monitor her father’s radiation dose levels, gave rise to violation of Article 2. The Court did not dispute that the respondent state was generally under a positive obligation to protect the right to life. Accordingly, it defined its task as determining whether ‘given the circumstances of the case, the state did all that could have been required of it to prevent the applicant’s life from being avoidably put at risk.’ The Court could not, however, establish a connection between any omission by the state and the disease from which the particular applicant suffered: ‘[I]t is clearly uncertain whether monitoring of the applicant’s health in utero and from birth would have led to earlier diagnosis and medical intervention such as to diminish the severity of her disease.’23Botta v Italy is another case where the absence of proximity was decisive. The applicant complained that the state had failed to take measures to remedy omissions imputable to private bathing establishments, which prevented disabled people from gaining access to the beach and the sea. The Court found that there was no ‘direct and immediate link’ between the measures sought by the applicant and his rights under Article 8 of the ECHR.24 Causality and its function in the case law of the Court have not received adequate consideration in the existing literature. Benedetto Conforti, a former judge of the Court, is among the few who have noted its importance: A conclusion that can be drawn from the Strasbourg case-law is that no violation is found in cases where there is lack of a causal link between the behaviour of the State and the event. The same reasoning applies to cases where the event is not ‘the immediate consequences of state behaviour’, i.e. where the State omission is too remote in the chain of the various circumstances which led to the final wrongful event. … Couched in simple terms, in the opinion of the present author, the causation, as it appears in the case-law of the Court concerning the obligations of prevention, is strictly linked to the ‘foreseeability’ of the wrongful event. It is an ex post test of ‘foreseeability’ of the event: even if the event was predictable there is still room, even after the wrongful event occurred, for verifying its place in the chain of events.25 Conforti correctly draws attention to the interdependence between foreseeability of the harm and the imputability of the harm to an omission by the state. States are certainly not omniscient entities and it cannot be expected from them to know about and foresee any potential harm that individuals might suffer. It must therefore be established that the state knew or ought to have known about the harm.26 Accordingly, the establishment of causation is related to the establishment of knowledge.27 3. KNOWLEDGE AND CAUSATION It is important to clarity two points in relation to the knowledge requirement. First, it is not necessary to prove actual knowledge; it suffices to demonstrate that the state should have known about the harm.28 Secondly, the knowledge requirement operates differently in different contexts. In some circumstances, knowledge that a specific individual might be harmed has to be established.29 In other circumstances, however, this will be too demanding, since the applicant might simply be a representative victim of some general failure in the national regulatory system.30 In these circumstances, it suffices to establish that the state knew or ought to have known that this system caused harm. It is not necessary to show that the state knew or ought to have known that the particular applicant would be negatively affected.31 It might be enough to demonstrate that the state knew or ought to have known about the source of the harm.32 How is knowledge of an organisational entity such as the state established? The mere existence of legislation against certain harms suffices for concluding that the state knew about these harms.33 The Court has also referred to the standard of ‘objective scientific research’,34 to various reports that might have been drafted at national level, and to various channels of communication between state institutions.35 The state is to a certain extent involved in many activities, such as issuing permits,36 which might also facilitate the establishment of knowledge. As Lavrysen has argued, knowledge is not a static issue, and the state should constantly anticipate new harms.37 On the other hand, the existence of knowledge and its accuracy might be contingent on state resources, which warns against unreasonable expectations concerning the state’s capacity to augur potential harms. This unavoidably has an impact in assessing whether these harms bear nexus with the state. In this context, the Court has not yet fully appreciated the following distinction: whether the question as to when the authorities ‘ought to have known’ of the existence of a risk can be answered by asking whether (i) they ought to have correctly assessed the risk based on the information they would have had if they had carried out their duties with due diligence or, instead, (ii) they ought to have known of the risk based on the information that was in fact available to them. The first alternative presupposes that the authorities actually had duties, which might be a premature conclusion since it might be also contingent on the reasonableness of imposing such duties.38 This can explain the reluctance of the Court to make fine distinctions and instead to bundle issues of knowledge, reasonableness and causation in its assessment of state responsibility in the framework of positive obligations. The merging of these elements will resurface as an issue at various other points below. Knowledge and causation are intertwined elements also because in some circumstances the degree of foreseeability of the harm might be such that it is from that alone that the requisite proximity between the harm and the state can be deduced. Still, it is helpful to maintain the distinction between the different elements, since in other circumstances it would make little sense to enquire what measures could have prevented the harm, if the state authorities did not know about the risk of such harm in the first place.39 4. THE STANDARD OF CAUSATION What standard of causation has been applied in the case law? Ugrekhelidze, a former judge of the Court, has noted: It must be ascertained whether the violation would have been avoided had the relevant positive obligations been duly honoured, i.e. whether that is a causal connection between the failure to honour them and the violation. This is necessary in order for the attribution of international legal responsibility to a State to have a solid objective legal basis.40 Framing causation as requiring an affirmative answer to the question whether ‘violation would have been avoided’, as proposed in the above quotation, is however problematic for the following reason. It appears that Ugrekhelidze assumes that the cause of the harm is the state omission and the consequence of this omission is the harm. These strict causality links are an erroneous reflection of the requirement for proximity between the harm and the state conduct as developed in the case law. The omission by the state might be just one factor contributing to the occurrence of the harm. The ‘but for’ test, which means that but for the state failure the harm would not have happened, has been explicitly rejected by the ECtHR.41 From the perspective of the individual who claims to be a victim of a human rights violation, it would be too demanding and, ultimately, it might be impossible to prove that if the state had adopted effective protective measures, the abuse would not have happened.42 In short, there is no requirement that but for the omission, harm would not have materialized. Yet as highlighted in Section 2 above, there needs to be a proximity between the harm and the state omission, and the Court has formulated the following standard to this effect: ‘A failure to take reasonably available measures which could have had a real prospect of altering the outcome or mitigating the harm is sufficient to engage the responsibility of the State.’43 Avoidance of the harm and its mitigation are formulated as alternatives, which leads to further relaxation of the standard. In addition, the undertaking of protective measures by the state might only have had a real prospect of avoiding or mitigating the risk of harm, which adds further flexibility to the causation analysis. For example, in O’Keeffe v Ireland, the enquiry was framed as to whether ‘effective regulatory framework of protection in place before 1973 might “judged reasonably, have been expected to avoid, or at least, minimise the risk or the damage suffered” by the present applicant.’44 The Court does not consistently refer to the ‘real prospect’ test in all of its judgments, though. Note in this respect that the Court uses different expressions in order to refer to the causation between the harm and any omissions. For example, in Vilnes and Others v Norway it framed the question as to whether harm is ‘caused’, ‘attributable’ or ‘imputable to any specific shortcomings for which he [the applicant] criticized the State.’45 This question was asked without considering the degree of likelihood that the absence of these shortcomings would have mitigated the harm or the risk of harm. In Budayeva and Others v Russia, the review was framed as whether ‘there was a causal link’ between the serious administrative flaws that impeded the implementation of the land-planning and emergency relief policies and the death and the injuries sustained by the applicants.46 The Court has also used the expressions ‘direct causal link’,47 ‘direct and immediate link’48 and ‘strong enough link’.49 The term ‘nexus’ has been also utilized: ‘[t]he combination of these factors shows a sufficient nexus between the pollutant emissions and the State to raise an issue of the State’s positive obligation under Article 8 of the Convention.’50 The expressions ‘due to’51 and ‘can be linked directly’ have also been used.52 In E. and Others v United Kingdom, the Court referred to the standard of ‘significant influence’: the failings of the relevant authorities disclosed in the case ‘must be regarded as having had a significant influence on the course of events.’53 In contrast, in Makaratzis v Greece, it was observed that the deficiencies in the legal and administrative framework had a ‘bearing … on the way in which the potentially lethal police operation culminating in the applicant’s arrest was conducted.’54 In Hiller v Austria, the Court found that ‘M.K.’s escape [from psychiatric hospital] and subsequent suicide had not been foreseeable for the hospital and was not therefore attributable to it;’ this is a formulation that echoes the above discussion about the interdependence between foreseeability and causation.55 It is doubtful whether these various terms reflect any differences in the substance of the analysis, varying scrutiny as to the proximity element and different standards of causation. The Court has not developed anything close to a consistent terminology. Overall then and despite the usage of the ‘real prospect’ standard that appears to offer a more principled approach, uncertainty pervades the case law. 5. CONTROL AND CAUSATION A. The Rules on Attribution Against the backdrop of the above tangle and intricacy in the case law, the question emerges whether it is possible to find some structure by drawing inspiration from other areas of the law on state responsibility. Within the framework of state responsibility, issues of proximity arise in contexts other than positive obligations.56 More specifically, the rules of attribution in general international law themselves reflect lines of proximity. Attribution as defined in the International Law Commission (ILC) Draft Articles reflects rules for connecting conduct to the state.57 The rationale behind these rules is ‘limiting responsibility to conduct which engages the State as an organization.’58 As I will show below, the role of proximity in the realm of positive obligations is very similar: limiting the responsibility of the state to circumstances where the state is engaged in the harm as an organization. It needs to be acknowledged, however, that once attribution is established, the causation between state action and harm is evident. In contrast, in the context of positive obligations and claimed omissions, the lines of causation might not be that easily discernible and might raise challenges. The rules on attribution connect agents and entities to the state. Conduct is attributable to the state when committed by its actual organs59 and de facto organs60 or by entities directed and controlled by the state.61 As to the first group of organs, the fundamental principle is that the state is responsible for the actions of its organs, even if they act ultra vires62 and even if they are no longer under the control of the state.63 That the state controls its organs is a normative assumption; the capacity to control and the actual control are irrelevant. The conduct of state organs thus might give rise to per se responsibility on the part of the state.64 As the ILC has framed it: ‘[t]he attribution of conduct to the State as a subject of international law is based on criteria determined by international law and not on the mere recognition of a link of factual causality [emphasis added]’.65 With regard to non-state entities, their actions may be attributable to a state if they act under the instruction, direction or control of that state. Attribution arises in this context because ‘there exists a specific factual relationship between the person or entity engaging in the conduct and the State’ and ‘a real link between the person or group performing the act and the State machinery’.66 The ILC Draft Articles do not specify the level of control required, which has led to controversies. The International Court of Justice has applied the test of ‘effective control’; while other adjudicative bodies have endorsed the standard of ‘overall control.’67 This debate need not to detain us here. It might come as a surprise that I mention the rules on attribution in the context of positive obligations. While these rules are certainly of importance in the context of negative obligations, they are generally perceived as irrelevant when the case is formulated as one involving a failure to fulfill positive obligations.68 The obligation to ensure human rights does not require a determination that the actual harm is attributable to the state in the sense of the ILC Draft Articles on State Responsibility69 and the previously mentioned rules of attribution. The triggering of positive obligations and the scope of these obligations are contingent on the primary obligations at stake, which are not a subject of the law on state responsibility as such. However, it is still relevant to engage with attribution since the rules of attribution under international law articulate lines of proximity. They express relationships of directness and immediacy between the act of the state and the harm. It is meaningful to consider the justifications and the theoretical underpinnings of these relationships, so that we better understand the linkages between harm and state conduct in the form of omission. B. The Role of Control and the Extension of the Logic of the Rules on Attribution The rules on attribution seek to establish a nexus between the state and the agent who caused the harm. The status of the agent is thus of importance and the harm caused by him/her is directly attributable to the state. Even if the conduct is in breach of the national legislation or the state agent exceeds the authority granted by national law (ultra vires acts),70 attribution is still established as long as the organ acts within its capacity.71 This must be differentiated from cases ‘where the conduct is so removed from the scope of their official functions that it should be assimilated to that of private individuals, not attributable to the state.’72 In contrast, Article 8 of the ILC Draft Articles highlights the question of ‘direction and control.’ It clarifies that ‘such control will be attributable to the State only if it [the State] directed or controlled the specific operation and the conduct complaint of was an integral part of the operation.’ However, as observed by Evans, this is not the way the ECtHR has extended the scope of the obligations under the ECHR. It is rather positive obligations that have served this purpose.73 For example, there might be situations where it is not really possible to show that the state has directly caused harm through its agents or entities under its control.74 Two basic principles follow from the above. First, acts of state organs are attributable to the state and the issue of control seems to be immaterial.75 Secondly, when the first principle is not applicable since, for example, none of the requirements in Articles 4, 5 and 6 of the ILC Draft Articles is fulfilled, the second principle is triggered: control implies responsibility.76 This principle is not sidelined once a case is framed as involving positive obligations. On the contrary, control by the state (not in the sense of Article 8 of the ILC Draft Articles though, which aims to link the state with a specific agent) is still relevant for determining the scope of these positive obligations. Public authorities are established to fulfill prescribed aims and they are conferred powers. They assume control over areas of activity, in this way putting themselves in proximate relationship with harm that might arise in these areas.77 It follows that control structures lines of proximity. The more control, the closer proximity may be expected between state conduct and harm, and accordingly, the positive obligations are more demanding. These positive obligations are thus commensurate with the extent of the control. In this sense, we can see some extension of the logic that applies to the rules on attribution. This normative account fits current practice since, as we will see below, it is reflected in the case law under the ECHR. The question which emerges at this junction is what principles apply if the state authorities are out of control. Do states have positive obligations when they do not have control? Can states absolve themselves from responsibility under human rights law by simply saying that they did not have control? Can they decide to relinquish control and free themselves from responsibility? The presumption that operates is that states have control over their territory and therefore continue to be under the obligation to ensure the rights enshrined in the ECHR.78 This implies that States have to re-assert control and take measures to secure these rights as a matter of principle.79 This certainly does not translate into state responsibility for failure to ensure human rights in every concrete case. States might face some practical difficulties in re-asserting control and in the degree of control that they can practically exercise; however, these will have to be taken into account in the assessment of the scope of the positive obligations in the particular case. In addition to issues of practicality and feasibility, certain areas of activities are underpinned with the normative assumption that the state must wield more control. I engage with these areas in more detail in Section 5.D below. First, however, I turn to circumstances where state agents inflict harm and how the positive obligations in relation to this harm are shaped by the degree of control exercised. C. Control and Prevention of State-inflicted Harm Positive obligations are usually analysed in circumstances when private actors cause harm. However, they are as much relevant in circumstances where state agents inflict harm. States are equally under the obligation to structure the relationships between their agents and individuals in such a way that harm is prevented and, if it occurs, is adequately addressed. In this context, the harm caused by failures in these structures is more closely proximate to the state and, accordingly, the obligation to prevent it is more demanding. For these reasons and in relation to the right to life, the Court has observed that ‘[w]hen lethal force is used within a “policing operation” by the authorities, it is difficult to separate the State’s negative obligations under the Convention from its positive obligations’.80 Still, in its analysis the Court distinguishes the two, and to this effect it has been established that Article 2 requires careful scrutiny not only as to whether the use of force by state agents was strictly proportionate to the aim of protecting persons against unlawful violence (negative obligation),81 but also whether the overall operation was ‘planned and controlled by the authorities so as to minimize, to the greatest extent possible, recourse to lethal force’ (positive obligation).82 The intensity of the above positive obligation is influenced by the level of control that the state has over the situation. For example, in Giuliani and Gaggio v Italy, the absence of foreseeability as to the course of the events, as well as the ensuing reduced level of control by the state over the situation, were taken into account in determining whether the organization and the planning of the policing operations were compatible with the obligation to protect life.83 Similarly, in Isayeva v Russia, where the applicant’s close relatives were killed by indiscriminate bombing by the Russian military, one of the factors considered by the Court was that the military operation conducted by Russia was not spontaneous and, therefore, the state had control over the circumstances.84 Likewise, in Makaratzis v Greece the Court did not overlook the fact ‘the applicant was injured during an unplanned operation which gave rise to developments to which the police were called upon to react without prior preparation.’ In the latter judgment it was also added that if the unpredictability of the events and the resultant reduction of the level of control were not taken into account in the assessment of the positive obligations, this might lead to the imposition of an ‘impossible burden on the authorities’.85 In the course of assessing the control over the situation and the scope of the positive obligations, the reasons as to why the state authorities did not have control are also of importance. Absence of foreseeability could be a reason. In Makaratzis v Greece, however, it was established that the degeneration of the situation and the ensuring chaos was largely due to the fact that at the time neither individual police officers nor the chase, seen as a collective police operation, had the benefit of the appropriate structure which should have been provided by the domestic law and practice.86 Since the absence of control was seen as attributable to deficiencies in the national legislation, Greece was found to have failed to protect the right to life. Therefore, despite the unpredictability of the events and the ensuing reduced level of practical control over the circumstances, the normative expectation that police operations are regulated by laws was determinative. Finogenov and Others v Russia, a case involving hostage-taking in a theatre that was stormed by the Russian authorities after the injection of an unknown narcotic gas, strongly supports the argument that there is a correlation between the control and the scope of the positive obligations. In Finogenov and Others v Russia, the Court modified the well-established Osman test, by observing: The authorities’ positive obligations under Article 2 of the Convention are not unqualified: not every presumed threat to life obliges the authorities to take specific measures to avoid the risk. A duty to take specific measures arises only if the authorities knew or ought to have known at the time of the existence of a real and immediate risk to life and if the authorities retained a certain degree of control over the situation.87 The Court delineated a different intensity of obligations and applied a different degree of scrutiny depending on the degree of control that the state had over the situation. It distinguished Finogenov from the above mentioned Isayeva v Russia since ‘[t]he hostage taking came as a surprise for the authorities, so the military preparations for the storming had to be made very quickly and in full secrecy.’ In addition, the authorities were not in control of the situation inside the theatre where the hostages were kept, which was also of material consideration.88 In contrast to the storming of the theatre, however, the positive obligations in relation to the subsequent rescue operation were of different intensity since ‘no serious time constraint existed and the authorities were in control of the situation.’ This justified a different approach when assessing the conduct of the Russian authorities.89 While the use of the gas and the storming was found not to be disproportionate measures in breach of Article 2, the rescue and evacuation operation were found inadequate. The latter operation was subjected to more thorough scrutiny because it was not spontaneous. In addition, it could be expected from the authorities that they had some general emergency plan and some control of the situation outside the building where the rescue efforts took place. The predictability of the harm also implied a higher level of control and more demanding positive obligations: ‘the more predictable a hazard, the greater the obligation to protect against it.’90 In conclusion, control over the situation is crucial for a finding of proximity, and important for assessing the intensity of the positive obligation. At the same time and as already intimated in the end of Section 5.B above, positive obligations themselves might require more pervasive control in certain circumstances. The next section will expand on this argument. Prior to this, however, the circularity in the above argumentation needs to be acknowledged. If states have control they put themselves in closer proximate relations with harm that might materialize and the positive obligations invoked in these circumstances are more robust. At the same time, the positive obligations themselves might require more control by the state. Instead of occluding this paradox, it should be rather openly acknowledged. It is beyond the scope of this article to resolve it because it is part of a much broader issue that ultimately concerns the role of the state in the society and to what extent and under what circumstances this role should be more intrusive.91 D. Assumption of Control in the Area of Public Services A cluster of positive obligations cases involve provision of public services. These services can be provided by public bodies or by private bodies. It has been well-established in the case law that states are not absolved of their human rights law obligations by delegating certain services to private bodies.92 The designation of the body is thus not of significance. For example, the Court has observed that ‘in the public-health sphere, these positive obligations require the State to make regulations compelling hospitals, whether private or public, to adopt appropriate measures for the protection of patients’ lives.’93 The nature of the activity, however, that is, public services, is of important material consideration since the breadth of the positive obligation increases with a function of this nature. The normative assumption that operates in this context is that the state should assume control in relation to these services and, as a consequence, the intensity of the obligation rises. A question which emerges here is how to define public services.94 The importance of the interest at stake, and also the relational context, can be determinative in delineating the contours of this definition. For example, in relation to education, the Court has emphasized that children have no alternative but to attend school since primary education is obligatory.95 In O’Keeffe v Ireland, the Court went so far as to establish an ‘inherent obligation [of the State] to protect children in this context, of potential risks.’96 The applicant in this case complained that the system of primary education failed to protect her from sexual abuse by a teacher. In relation to this complaint, the Court held that the primary education context of the present case defines to a large extent the nature and the importance of this obligation. The Court’s case law makes it clear that the positive obligation of protection assumes particular importance in the context of the provision of an important public service such as primary education, school authorities being obliged to protect the health and well-being of pupils and, in particular, of young children who are especially vulnerable and are under the exclusive control of those authorities.97 Although the school in question was owned and managed by a non-state actor, the Court relied on the fact that education is an importance public service and on the vulnerability of children to determine the scope of the positive obligation.98 The importance of the activity in question and the control of the authorities were also emphasized in Ilbeyi Kemaloglu and Meriye Kemaloglu v Turkey, a case about the child who froze to death after being left alone by the school authorities in a heavy snow storm: [T]he State’s duty to safeguard the right to life is also applicable to school authorities, who carry an obligation to protect the health and well-being of pupils, in particular young children who are especially vulnerable and are under the exclusive control of the authorities.99 The same principle has been applied to psychiatric institutions, in relation to which ‘the State remained under a duty to exercise supervision and control’100 and nursing homes, which the state is under the obligation to regulate.101 Similar reasoning underpinned Iliya Petrov v Bulgaria, where a boy was severely harmed after entering a transformer and receiving an electric shock. The Court acknowledged that the boy was very unwise to enter such a dangerous place, but it still held that the ‘decisive factor’ for the occurrence of the incident was the deficient control by the authorities regarding the safety of electric transformers,102 the latter being a public service in relation to which the state is expected to exercise control. Cevriogli v Turkey is also illustrative in this respect. The case concerns the death of a boy as a result of falling into a large water-filled hole outside a private building under construction in a residential area. The Court referred to the ‘inherently hazardous nature’ of construction sites and, accordingly, the expectation that the state controls, inspects and supervises the activities at these sites.103 In its reasoning, the Court added that the state ‘in the present context had a more compelling responsibility towards the members of the public who had to live with the very real dangers posed by construction work on their doorsteps.’104 The above outlined expectation that in certain areas the state assumes control, shapes the approach to causation. For example, in Cevriogli v Turkey, it was admitted that no causal link may exist between the failings to inspect the construction site and the death of the boy ‘for the purposes of civil liability.’ However, in the context of state responsibility where the objective is to find the responsibility of a collective, strict causation lines are inappropriate. In this respect, the Court observed that ‘proper implementation of an inspection mechanism would undoubtedly have increased the possibility of identifying and remedying the failings which were responsible for the death of the applicant’s son.’105 E. Source of the Harm and the Related Level of Control The establishment of causation between failures on behalf of the state to take measures and harm is also affected by the source of the harm: whether the source is a natural or a man-made phenomenon. In both contexts, the state is under the general obligation to protect; however, in the event of a harm ensuing from a man-made phenomenon, proximity is easier to establish and the scope of the positive obligations is more demanding. In situations involving ‘dangerous activities’, where the harm is perceived as man-made, or in relation to events ‘regulated and controlled by the State’, it is easier to establish that state omissions are causative to harm. An example to this effect is Öneryildiz v Turkey, a case about an explosion at a garbage collection point. In its submissions to the Court, Turkey tried to challenge the extension of positive obligations under Article 2 to all circumstances of unintentional death. The Court, however, emphasized the dangerous nature of the activity and the ensuing expectation that the state regulates it. In this way, it established a proximity between the harm and failure by the state to regulate,106 which also implied that the positive obligation to ensure the right to life ‘must be construed as applying in the context of any activity, whether public or not, in which the right to life may be at stake, and a fortiori in the case of industrial activities, which by their very nature are dangerous, such as the operation of waste-collection sites.’107 In Vilnes and Others v Norway, the Court extended this logic to risky activities and observed that ‘it sees no need to consider in detail the degree of involvement of the respondent State in the hazardous activity in question, since the Convention obligation applies “any activity, whether public or not”’.108 The same approach was adopted in Kolyadenko and Others v Russia, a case about an urgent massive evacuation of water from reservoir, where the Court also emphasized that a reservoir is a man-made industrial facility.109 The above analysis is not modified by the nature of the agent that performs the activity. Whether that activity is performed by a public entity or by, for example, private corporations, is immaterial. Thus, the Court’s assertion in Brincat and Others v Malta, that the positive obligation to safeguard lives ‘may apply in cases, such as the present one, dealing with exposure to asbestos at a workplace which was run by a public corporation owned and controlled by the Government’ is confusing.110 If consistency in the case law is to be maintained, the public or private nature of the entity that engages in the activity is not pertinent. As the Court itself has observed, the positive obligation to protect life applies ‘in the context of any activity, whether public or not’.111 The above cases implicating man-made harm can be distinguished from circumstances where harm is caused by ‘natural’ disasters. It is more difficult to establish proximity and therefore the positive obligations are not that extensive in cases of natural disasters ‘which are as such beyond human control’ and ‘do not call for the same extent of State involvement’.112 In Budayeva and Others v Russia, a case about a mudslide causing loss of life and destruction of property, the Court observed that ‘[t]he scope of the positive obligations imputable to the State in the particular circumstances would depend on the origin of the threat and the extent to which one or the other risk is susceptible to mitigation.’113 It was added that the consideration of not imposing a disproportionate burden on the state ‘must be afforded even greater weight in the sphere of emergency relief in relation to a meteorological event, which is as such beyond human control, than in the sphere of dangerous activities of a man-made nature.’114 Still, against the background of the national authorities’ omissions in the implementation of land-planning and emergency relief policies in the hazardous area where the mudslide occurred, and the existence of a causal link between these failures and death of the victim, Russia was found to have failed to discharge its positive obligations under Article 2.115 The same logic has been extended in relation to the positive obligations corresponding to the right to property. In Öneryildiz v Turkey, the Court observed: Genuine, effective exercise of the right protected by that provision [Article 1, Protocol 1] does not depend merely on the State’s duty not to interfere, but may require positive measures of protection, particularly where there is a direct link between the measures an applicant may legitimately expect from the authorities and his effective enjoyment of his possessions.116 The establishment of this causal link is affected by the source of the harm. In Öneryildiz v Turkey such a link was found to be present: ‘there is no doubt that the causal link established between the gross negligence attributable to the State and the loss of human lives also applies to the engulfment of the applicant’s house.’117 In Budayeva and Others v Russia, however, the Court introduced a distinction along the following lines. As opposed to Öneryildiz v Turkey, where the events occurred ‘under the responsibility of the public authorities’ and where ‘[t]reatment of waste … is regulated and controlled by the State, which brings accidents in the sphere within its responsibility’,118 in Budayeva and Others v Russia the Court considers that natural disasters, which are as such beyond human control, do not call for the same extent of State involvement. Accordingly, its positive obligations as regards the protection of property from weather hazards do not necessarily extend as far in the sphere of dangerous activities of a man-made nature.119 While the implementation of certain measures (mud-defence infrastructure and an early warning system) were considered by the Court as vital for the protection of life in case of a mudslide (and in this sense the Court was prepared to find causality between the absence of these measures and the loss of life), no evidence was found that these measures could have prevented the damage to the applicants’ possession: ‘[I]t cannot be said that the causal link between the State’s failure to take these measures and the extent of the material damage is similarly well-established.’120 In Budayeva and Others v Russia, the Court thus appears to suggest that to a certain extent some of the damage to property was imputable to state negligence; however, it is difficult to ascertain how much. The Court’s approach is also based on the premise that even if mud-defence infrastructure had been installed, damage to property would still have occurred. F. Assumption of Control over the Victim121 While the preceding sections addressed the issue of state control over certain circumstances and activities, this section focuses on control over individuals. Once the state has undertaken any special responsibilities in relation to certain individuals, the lines of proximity between harm and state omissions solidify. States owe more extensive positive obligations to those with whom they have special ties. The primary example in this respect are prisoners, who are placed under extensive state control.122 This control implies more demanding positive obligations. These have been considered in various contexts (protection from private violence,123 protection from suicide,124 acceptable detention conditions, including prevention of health hazards,125 and provision of minimum socio-economic assistance). It will suffice to compare the scope of the positive obligations in relation to the provision of health care for prisoners with the scope of the positive obligation in relation to provision of healthcare to the population at large.126 In the latter context, the Court applies a high definitional threshold to find violations of Articles 3 and 8 in the sphere of socio-economic assistance.127 The Court has not excluded the possibility that violations of Articles 3 and 8 might be found in circumstances characterized with deprivations resulting from, for example, insufficient welfare benefits; however, such a finding will be made under exceptional circumstances.128 The exceptionality approach, however, will be modified if other elements intervene. As mentioned above, no such approach is applied in relation to prisoners. The Court has taken this very far since not even financial considerations (for example, the argument that the state does not have enough money to maintain prisons) are accepted.129 There are other circumstances where, due to the special relationship between the state and the victim and more specifically due to the assumption of control by the state over the victim, the logic of exceptionality is displaced. In Denis Vasilyev v Russia, the Court observed that the duty to protect is not ‘confined to the specific context of the military and penitentiary facilities.’130 It added: It also becomes relevant in other situations in which the physical well-being of individuals is dependent, to a decisive extent, on the actions by the authorities, who are legally required to take measures within the scope of their powers which might have been necessary to avoid the risk of damage to life and limb.131 The factual substratum of Denis Vasilyev v Russia was underpinned by omissions by police officers. In particular, after finding the applicant unconscious on the street, they left without calling for medical assistance. Despite the important assertion framed by the Court in the above quotation to the effect that the obligation to protect is triggered when the well-being of the individual is dependent on the state, the reasoning in the judgment overall is confusing. The Court pointed to various factors (vulnerability of the person, knowledge about his position, control over him once the state authorities knew about his position and the requirements under the national legislation to render assistance) to find that the state was in breach of its positive obligations. Overall, the reasoning manifests a heavy emphasis on the requisites of the national legislation to assist the person, which tilted the Court to find a violation of Article 3. In the next section, I will discuss in more detail the role of domestic legality. Here it is pertinent to observe that there might be other situations where the above mentioned exceptionality might be displaced. One such example is where the state has precluded the availability of alternative means of protection and assistance. Such a situation might transpire in relation to asylum-seekers, who might not be allowed to work in the first place, which will inevitably modify the analysis as to whether the absence of socio-economic assistance by the state causes harm falling within the scope of Article 3 or 8.132 Because of the special relationship between the person and the state that arises in the above-mentioned contexts, the foreseeability of the harm requirement might be loosened.133 In light of the special position of the victim in relation to the state, it might be hardly possible to argue that the state did not know or ought not to have known about the harm or the risk of harm. Some aspects of the case law, however, cause confusion. For example, in Nencheva v Bulgaria, the Court extensively reviewed whether the central national authorities knew about the dire circumstances of the disabled children who were accommodated in a home. In light of the fact that the home in question was under the control of these authorities, one is left to wonder whether the authorities should in any case have known about the risks faced by the children. Even if they did not actually know, there should have been mechanisms for channeling such information.134 In sum, the case law is ambiguous as to which factor has a dominating role: the actual knowledge about harm, on the one hand, or the normative supposition that state authorities should know or necessarily knew about the harm (or risk of harm) to individuals under their control. One final observation is due in this section. The special position of the victim can be related to the particular vulnerability of certain categories of persons.135 The Court has recognized various groups as vulnerable: children,136 asylum-seekers,137 prisoners, military conscripts,138 persons with disabilities,139 victims of domestic violence140 and Roma.141 It is useful, however, to distinguish between the sources of these various vulnerabilities. Some of them might be innate and inherent vulnerabilities (for example, children). Others might be related to the social context (for example, Roma, religious minorities).142 In this section, I draw attention to a distinctive vulnerability that stems from the specific relationship with the state and, in particular, from the exposure of the person to state power. Prisoners, for example, who due to their detention are exposed to state power and deprived of other sources of help, are placed in a special relationship with the state, which explains the more demanding positive obligations.143 Certainly, this special relationship could be intimately related to or substantiated by vulnerability stemming from innateness or social context (for example, protection of children in the context of compulsory education) and in this sense, the different sources of vulnerability might be interrelated. 6. Techniques for Avoiding Causation So far this article has covered areas where traditionally the state has had an important role and in this sense the state more easily places itself in closer proximate relations with harm that might materialize (for example, policing operations, provision of public services, industrial activities, restraints imposed upon individuals). In other areas, however, the intensity of the involvement of the state can be more controversial (for example, protection of the environment, regulation of private companies) and, as already suggested at the end of Section 5.C, the question as to how intrusive the role of the state should be and how much control the state should exercise can be more contentious. Moreover, the establishment of causation between harm and state omissions might be hampered by empirical and epistemological uncertainties. Since the Court might not want to see itself deeply implicated in normative judgments about the role of the state in society more generally, and since the Court might not be in a position to resolve empirical and epistemological uncertainties, it has crafted techniques to avoid making these judgments and resolutions. These techniques also imply that the Court does not have to directly confront the issue as to whether state omission is causative to harm. Two such techniques will be discussed here: domestic legality and procedural protection. A. Domestic Legality When an omission is contrary to the national regulatory framework, proximity between this omission and the harm amounting to violation of human rights law is easier to assume. In other words, when the national legislation or applicable regulatory standards themselves envisioned the undertaking of certain measures and these were not performed, the Court is more willing to accept that there is a nexus between the non-performance and harm. The underlying assumption is that the national regulatory framework was adopted in order to prevent harm. Once this is transposed at the level of the ECHR, whether or not this is indeed the case (to wit, whether or not generally or in relation to the particular applicant the proper application of the legal framework would have indeed prevented harm or reduced the risk of harm), seems to be less relevant, since the above assumption continues to operate. How non-compliance with the national legal requirements renders the proximity standard less stringent from the perspective of the applicant was made obvious in I v Finland. The applicant complained that a hospital has failed to guarantee the security of her data against unauthorized access. She worked as a nurse and was diagnosed as HIV-positive. At a certain point, she suspected that her colleagues were aware of her illness and soon her contract was not renewed. On the facts, it was not possible to determine whether her records were actually accessed by an unauthorized third person. As a consequence, it was not possible to prove a causal connection between deficiencies in the access rules in this particular hospital (that is, not maintaining a log of all persons who had accessed her medical files) and the harm that she has experienced (that is, dissemination of information about her medical condition). However, the Court did not find it necessary to prove such causation. Instead, it observed that ‘what is decisive is that the records system in place in the hospital was clearly not in accordance with the legal requirements’ in the national legislation and did not hesitate to find a failure on behalf of the state to fulfill its positive obligation under Article 8.144 The requirements laid down in the national legislation, including those that require positive measures, establish a baseline. Any deviation from this baseline is suspect and makes it easier to argue that a state’s failure to comply with its own baseline is causative of harm. Support for this approach can be found in the context of the right to life and the adequate standard of health care.145Lopes de Sousa Fernandes v Portugal and Elena Cojocaru v Romania are judgments in point. The Court observed that it would not speculate on the particular patient’s prospects for survival, if the measures required by the medical protocols had been actually undertaken. It sufficed for the establishment of state responsibility that there was ‘apparent lack of coordination of the medical services and … delay in administering the appropriate emergency treatment,’ which ‘attest to a dysfunctionality of the public hospital services’.146 In the reasoning of the majority, this dysfunctionality could be related to non-observance of medical protocols. On the facts, however, it was not entirely clear whether non-compliance with the protocols caused the death of the applicants’ relatives. The Court’s leniency regarding causality in the above-mentioned two cases prompted Judges Sajó and Tsotsoria to attach a Dissenting Opinion to the Chamber judgment in Lopes de Sousa Fernandes, in which they observed that it is hard to understand ‘how an alleged organisational negligence that did not result in death can be construed as the basis of State responsibility for failing to protect life.’ In their Joint Dissenting Opinion, they also added that even if there had been a causal relation, this is still not enough to find a violation. I will return to this aspect of their argument in Section 7.C below, where I address the distinction between incidental and systemic failures. Here is it pertinent to observe that the Grand Chamber in Lopes de Sousa Fernandes addressed the dissenters' concerns by observing that the question whether there has been a failure by the State in its regulatory duties calls for a concrete assessment of the alleged deficiencies rather than an abstract one. … Therefore, the mere fact that the regulatory framework may be deficient in some respect is not sufficient in itself to raise an issue under Article 2 of the Convention. It must be shown to have operated to the patient's detriment.147 It follows that the Grand Chamber has modified the Chamber's approach. As the above quotation shows, causality needs to be established between the organisational negligence and the condition of the specific applicant. Using the national regulatory framework as a metric has been applied in other areas of the case law. More specifically, in A v Croatia, a domestic violence case, the failure by the national authorities to implement measures ordered by the national courts was highlighted, and led the Court to conclude that the respondent state failed to ensure the victim’s right to private life.148 In Taskin and Others v Turkey, the applicant complained that the operation of a gold mine posed risks to their right to life and private life. The respondent government challenged the assertion that the operation of the mine had harmful effects, and in this sense scientific uncertainty permeated the facts. The Court did not find it necessary to engage with the issue as to whether the operation of the mine was indeed contributory to harm, since this operation was contrary to domestic law.149 Certainly, the state might have perfectly complied with the existing national legal and regulatory framework and still have failed to fulfill its positive obligations under the ECHR, since deficiencies in this very framework might be causative to harm.150 On the other hand, even if a failure to take certain protective measures was contrary to the national law and regulation, this is not in itself conclusive that the state has failed to fulfill its positive obligations under the ECHR. The Court might pursue further enquiries into lines of proximity.151 Domestic legality is thus not a conclusive test.152 This renders the analysis distinctive in comparison with cases framed as implicating negative obligations, where if a measure restricting the right is not ‘in accordance with the law’, this automatically renders the restriction contrary to human rights law.153 Still, as clarified above, even in the context of positive obligations, domestic legality plays an important role in shaping lines of proximity. In addition, the notion of legality does not only have a substantive aspect, which, as explicated above, concerns compliance with domestic regulatory frameworks; it also has a procedural aspect, to which I now turn. B. Procedural Protection In various fields of its case law on positive obligations, the ECtHR has added a procedural layer to the scope of the Convention rights by requiring states to ensure the availability of effective national procedures. As Brems and Lavrysen have observed, the rationale behind this move is the proposition that procedural guarantees are instrumental for better protection of the substantive guarantees.154 The task of this section is not to offer a general analysis of the development of procedural protection by the Court, which has been already done by the above-mentioned authors. Its focus is rather the underlying causality, namely the understanding that the substantive harm is less likely to have materialized, had the decision-making process at national level been of sufficient quality. The main argument that I advance here is that procedural protection offers an avenue for the Court to deal (or rather not to deal) with empirical and epistemological uncertainty. This uncertainty poses challenges in ascertaining the remoteness or the proximity between the harm sustained by the individual and state omissions; procedural protection resolves the difficulty. The Court can eschew conclusive determinations that certain substantive omissions cause harm, and instead can focus on procedural omissions and deficiencies at national level. The inclusion of a right to access to information in the environment-related judgments delivered by the Court supports the above argument.155 In Roche v United Kingdom, there was uncertainty whether the applicant had been put at risk through his participation in chemical tests. This uncertainty did not have to be resolved because the Court framed its task as considering whether a positive obligation arose to provide an “effective and accessible procedure” enabling the applicant to have access to “all relevant and appropriate information” which would allow him to assess any risk to which he had been exposed during his participation in the tests.156 The Court also added that the applicant’s uncertainty ‘as to whether or not he had been put at risk through his participation in the test carried out at Porton Down could reasonably be accepted to have caused him substantial anxiety and stress.’157 Accordingly, the core issue in the case was shaped as not how the tests themselves contributed to the harm falling within the scope of Article 8; but how the denial of access to information about the tests caused harm falling within the scope of Article 8. These are certainly two distinct, although related, types of harm. Access to information is only one element for assessing procedural protection. As Hatton and Others v United Kingdom shows, the Court can scrutinize more generally the quality of the national decision making process.158 In particular, scientific uncertainty underpinned the case since ‘[t]he position concerning research into sleep disturbance and night flights is far from static’.159 No finding on this specific point had to be made at the level of the Strasbourg Court because the national decision-making process was found adequate. No fundamental procedural flaws in the preparation of the night flight scheme at Heathrow airport were found. Vilnes and Others v Norway is an example of a case where quite wide-ranging assumptions about causal relationships and procedural protection were made.160 This merits more detailed examination. The applicants, who worked as divers in the North Sea for private companies, complained that Norway did not adopt an effective legal framework of safety regulations that could prevent the divers’ lives and health being put at risk. While dismissing most of the divers’ allegations, the ECtHR still found a failure on behalf of Norway. The underlying reason was that the companies were left with little accountability vis-à-vis the state authorities in relation to the usage of diving tables, which were treated as company business secrets.161 In other words, Norway allowed a situation in which the divers were not informed about the health and life-related risks pertaining to the usage of diving tables. There are two types of causality underpinning Vilnes and Others v Norway that have to be separated and further clarified. The first relates to the extent to which the rapid decompression tables did in fact contribute to the applicants’ medical problems. The standard for assessing this contribution was formulated in the following way by the Court: The Court, having regard to the parties’ arguments in the light of the material submitted, finds a strong likelihood that the applicants’ health had significantly deteriorated as a result of decompression sickness, amongst other factors. This state of affairs had presumably been caused by the use of too-rapid decompression tables. … Thus, with the hindsight at least, it seems probable that had the authorities intervened to forestall the use of rapid decompression tables earlier, they would have succeeded in removing more rapidly what appears to have been a major cause of excessive risk to the applicants’ safety and health in the present case.162 However, as the Court framed the case, the core issue was not that the state had not prevented the use of rapid decompression tables as such or that the state had not prevented their use earlier. As the Court alluded in the above citation, the conclusion that elimination of the tables would have reduced the risk can be reached only with the benefit of hindsight. At the material time in the past, it was widely believed that diving did not have serious long-term effects in the absence of decompression sickness. A procedural duty was rather placed at the heart of the case, namely the duty of the state to provide information essential for the divers to assess the health risks. Although the risks associated with diving at that time were still disputed, the raison d’etre of the decompression tables themselves was to provide information essential for the assessment of risk to personal health.163 As the Court observed, the authorities were aware that the diving companies kept the tables confidential for competitive reasons.164 The authorities failed to enlighten the divers about the risks, which in the logic of the judgment would have enabled the divers to give informed consent to the taking of such risks.165 In light of the above, a second type of causality was brought forward in the judgment: if the state had ensured that the divers could assess the risks to their health and give informed consent to the risks involved, this would lead to the elimination of the use of the rapid tables. The Court frames this in the following way: Had they done so [had the authorities ensured that the companies provide information by not keeping the tables confidential] they might conceivably have helped to eliminate sooner the use of rapid tables as a means for companies to promote their own commercial interest, potentially adding to the risks to divers’ health and safety.166 Two comments are pertinent here. First, the standard of ‘conceivably have helped’ as a test of causation appears to be very low, and has not been used before Vilnes and Others v Norway. Secondly, the reasoning in the above quotation leads to contradictions within the judgment. In light of the absence of clear expert understanding at the material time of the consequences of the usage of decompression tables, as in fact acknowledged by the Court itself, it is hard to maintain that if the divers themselves had had access to the tables, this information would have been of use and led them to change their behaviour. The above analysis is not a suggestion that the state was not at fault. The wrong that can be clearly imputed to the state is that it allowed the companies to treat the decompression tables as their business secret and keep them confidential for competitive reasons. How that wrong can be linked with the harm sustained by the particular applicants so that this harm can be translated into international responsibility under the ECHR, however, is a separate question. In its response to this question, the Court appeared to be very flexible when assessing causation. Admittedly, this could be counterbalanced with the understanding that it is not reasonable to allow business secrets in relation to issues that raise controversies as to their impact on human health. In other words, it would not have been anything close to an unreasonable burden on the state to demand disclosure of the tables. It is precisely the test of reasonableness that the next section will address. 7. Technique for Limiting Responsibility when Causation is Present A. Reasonableness The Court has consistently reiterated that ‘the positive obligation to protect is to be interpreted in such a way as not to impose an excessive burden on the authorities.’167 The scope of the positive obligations under the Convention cannot therefore be unreasonable. When the Court refers to reasonableness in the context of positive obligations, it has in mind public interests—including public policy considerations, budgetary concerns and the rights of others—as factors that might compete with the assistance and protection interests of the particular applicant.168 This section considers how the reasonableness standard might affect and mould the assessment of causation. As already suggested above in relation to Vilnes and Others v Norway, even if the lines of causation between state omissions and harm are tenuous, reasonableness might intervene and buttress a finding of a violation. Here I will focus on the other end of the spectrum, where even if there is clear factual causality between state conduct and harm, additional factors related to the reasonableness standard influence the determination whether the respondent state is responsible under human rights law. Mastromatteo v Italy is illustrative in this respect. A brief summary of the facts is apposite here. The applicant’s son was murdered by a gang of criminals. The murder was carried out at a time when they were on special prison leave or benefiting from a regime of semi-liberty. It was undisputed that if the state had not released the criminals, Mastromatteo would not have been murdered. In this sense, one can see a clear causation: failure by the state to keep them in prison resulted in severe harm. The Court, however, observed that a mere condition sine qua non does not suffice to engage the responsibility of the State under the Convention; it must be shown that the death of A. Mastromatteo resulted from a failure on the part of the national authorities to ‘do all that could reasonably be expected of them to avoid a real and immediate risk to life of which they had or ought to have had knowledge’, the relevant risk in the present case being a risk to life for members of the public at large rather than for one or more identified individuals.169 The Court went on to examine the decision of the national authorities to let the criminals on leave and concluded that there was nothing in the material before the national authorities to alert them to the fact that the release of M.R. or G.M. would pose a real and immediate threat to life, still less that it would lead to the tragic death of A. Mastromatteo as a result of the chance sequence of events which occurred in the present case.170 In light of the benefits associated with rehabilitation programs for prisoners, the judgment suggests that it would not have been reasonable to keep M.R. and G.M. in prison when the state did not know that they posed ‘real and immediate’ risk to harm. The Court thus took note of broader considerations related to reasonableness and the interest of others,171 namely the benefits associated with letting prisoners on leave for the purpose of social reintegration. B. The ‘Real and Immediate Risk’ Standard A question that merits further investigation at this junction relates to the role of the ‘real and immediate risk’ test and, more specifically, how it affects causation and the reasonableness standard. In support of the standards set in the above-quoted paragraphs from Mastromatteo v Italy, the Court referred to Osman v United Kingdom. The test of ‘real and immediate risk to life’ was established therein in the context of the positive obligation to take protective operational measures. For a violation of this obligation to be found it must be established to its [the Court’s] satisfaction 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 or individuals from the criminal acts of a third party and that they failed to take measures within the scope of their powers which, judged reasonably, might have been expected to avoid that risk.172 This test is relevant in circumstances where the state knew or should have known about dangers for a specific identifiable individual from the criminal acts of another individual.173 The Court has never clarified the meaning of ‘real’ and ‘immediate’ and usually bundles these criteria together without elucidating how they differ. Often, it refers to them as ‘general principles’ without clarifying their application to the concrete circumstances of the case. ‘Real’ risk could be understood as risk that is objectively given and not merely speculative. The qualifier ‘real’ could refer to the likelihood that the risk will materialise.174 In this sense, a parallel can be drawn with the ‘real prospect’ test discussed in Section 4, which refers to the likelihood that a protective measures could obviate the risk. ‘Immediate risk’ could be understood as risk that is ‘present and continuing’.175 It has also been suggested that immediacy implies that the harm could be expected to ‘materialize at any time’.176 The immediacy prong of the test and the identifiability of the person exposed to the immediate harm are the distinguishing features that characterize the circumstances where the positive obligation of taking protective operational measures is applicable. The existence of a ‘real and immediate risk’ needs to be conceptually separated from the existence of official knowledge about the risk (the latter requirement was addressed in Section 3 above). Such a risk might exist even if the state authorities are not aware of it, and thus state responsibility will be precluded when there is no official awareness. However, this preclusion of responsibility is undermined by the requirement formulated as ‘ought to have known’ about the risk, which is given as an alternative to ‘the authorities knew’ standard. As already observed in Section 3 above, the ‘ought to have known’ test implies calculations as to how reasonable it is to actually expect that the authorities know about and foresee harm. It is important to clarify here that the ‘real and immediate risk’ test is not of general applicability to positive obligations. Accordingly, McBride’s argument that the Court has responded to the difficulties arising from different possible sources of harm by ‘adopting an exacting causal test, namely that the risk of death from the supposed threat be “real and immediate”’ needs to be rejected.177 This test is rather limited to circumstances where the person exposed to the harm can be individualized and identified in advance, and the risk of harm is immediate. In this sense, Osman v United Kingdom and the other cases in which the positive obligation of taking protective operational measures has been found to be triggered,178 have a factual substratum that only tenuously resembles Mastromatteo v Italy. The ‘real and immediate risk’ test might appear well-suited in cases where identifiable individuals are exposed to the risk of criminal acts of others, since it might be unreasonable to expect the authorities to take protective operation measures of an ad hoc nature without some imminence and concreteness of the risk to which a particular individual is exposed. In other words, it would be unreasonable to expect the state to take such measures in relation to an indeterminate class of people. In contrast, when the issue is general protection of the society, the test of ‘real and immediate risk’ is ill-suited. Therefore, the factual substratum of Mastromatteo v Italy does not call for the invocation of protective operational measures since there was no immediate danger to the life of any specific individual; it rather reveals circumstances where the applicant’s son happened to be an accidental victim of various errors in relation to the application of the national regulations about prison leave.179 These errors were, however, assessed as not sufficiently material and systemic, an issue I address in the next subsection. In sum, ‘the real and immediate risk’ test limits the circumstances where the state might be found responsible for its omissions. However, this limitative function is not of general applicability.180 It is not appropriate to operate in circumstances demanding general protection of the society, where there is no immediate danger to any specific individual,181 and where the state is rather under the obligation to put in place a general legislative and administrative framework for regulating activities so that harm is prevented.182 The current uncertainty as to the meaning of ‘immediate’ risk, however, and the threshold of immediacy required, will continue to spread confusion in the case law.183 The reason for this confusion is not simply the Court’s inadequate stringency in the application of the standards, a point already intimated above; there are also broader considerations that need to be seriously considered. In particular, if the state were to take protective actions against any potential risk regardless of its immediacy, then we might be confronted with the problem of a too intrusive state. This is a dilemma already identified in Section 5 above and, it calls for caution when expanding the scope of the positive obligation of putting in place general legislative and administrative framework for regulating activities so that harm is prevented. On a related point, we as a society might have to accept certain levels of risk,184 and this might militate against expansive construction of positive obligations. This is an argument that bulwarks the proposition that even in cases of clear factual causality, violations of human rights should not be found because society has to tolerate and accept certain risks. C. Incidental as Opposed to Systemic Failures In addition to the ‘real and immediate risk’ test, another technique utilized by the Court for limiting state responsibility relates to the distinction between incidental and systemic failures. As touched on above, a reason that might explain the outcome in Mastromatteo v Italy, where no violation was eventually found, is that the errors on the part of the national authorities appeared incidental and not systematic.185 In this sense, it might be unreasonable to find Italy responsible under ECHR for isolated errors and omissions even if they might have caused harm. This line of reasoning clearly emerges in the context of Article 2 and health care: [W]here a Contacting State had made adequate provision to secure high professional standards among health professionals and to protect the lives of patients, it cannot accept that matters such as error of judgment on the part of a health professional or negligent coordination among health professionals in the treatment of a particular patient are sufficient of themselves to call a Contracting State to account from the standpoint of its positive obligations under Article 2 of the Convention to protect life.186 As pointed out in the Joint Dissenting Opinion of the Chamber judgment in Lopes De Sousa Fernandes v Portugal, the majority had set a standard that casual acts of negligence by members of staff would not give rise to a substantive breach of Article 2. The Grand Chamber judgment in Lopes De Sousa Fernandes v Portugal, not only unequivocally upheld the principle embodied in the above quotation, but it added an additional layer of restrictiveness as compared to the approach by the Chamber in the same case.187 Any deficiencies that might give rise to a substantive violation of Article 2 in medical cases, need not only be systemic or structural (and thus not ‘a mere error or medical negligence’), but they must also implicate denial of immediate emergency care.188 The suggestion that incidental failures might not afford a basis for state responsibility is reminiscent of another test that has been invoked in the case law, namely the ‘significant flaw’ test. The Chamber invoked this test in precluding responsibility in Söderman v Sweden. In particular, it observed that ‘only significant flaws in legislation and practice, and their application, would amount to a breach of a State’s positive obligations under the said provision [Article 8]’.189 Subsequently, however, when the Grand Chamber reviewed Söderman v Sweden, it partially rejected the ‘significant flaws’ test: [S]uch a significant-flaw test, while understandable in the context of investigations, has no meaningful role in an assessment as to whether the respondent State had in place an adequate legal framework [emphasis in the original] in compliance with its positive obligations under Article 8 of the Convention since the issue before the Court concerns the question whether the law afforded an acceptable level of protection to the applicant in the circumstances.190 ‘Acceptable level of protection’ is a standard the Court had never used before.191 What is also interesting about Söderman v Sweden is that no reasonableness standard was even invoked: the Grand Chamber was simply not satisfied that the relevant Swedish law, both criminal and civil, as it stood at the time when the applicant’s stepfather covertly attempted to film her naked in their bathroom for a sexual purpose, ensured protection of her right to respect for her private life.192 In particular, the act of filming was not criminalized. The approach by the Grand Chamber in Söderman v Sweden barely squares with the approach in Mastromatteo v Italy, or in fact, in other judgments where the issue was whether the national legal and administrative framework was effective and where reasonableness was applied as a factor in the delimitation of the scope of the positive obligation.193 Can any explanations concerning this inconsistency be found? An overview of the case law shows that when criminal legislation is invoked as a means of ensuring the rights,194 no test of reasonableness is applied and, in fact, the issue of causation seems to be immaterial. As a consequence, it is simply assumed that criminalization (or an interpretation of the national criminal law so that its reach is more expansive) contributes to better protection of human rights.195 In contrast, when legal frameworks other than criminal law are at issue, reasonableness and competing interests are included as factors in the analysis. Mastromatteo v Italy exemplifies this since ultimately the issue in this case was whether the national framework regulating prison leave contained sufficient safeguards to protect the general population from prisoners on leave. The approach described above, under which the test of reasonableness is not applied when the issue is whether the national criminal law ensures effective protection, is, however, balanced in the following way. Criminalization, as a means of ensuring human rights, is required only where the harm sustained by the victim meets a certain threshold of severity. This will be the case, for example, where Articles 2, 3 or 4 are found applicable,196 and where ‘fundamental values and essential aspects of private life are at stake.’197 In sum, it was relatively easy to reject the ‘significant flaw’ test in Söderman v Sweden because the effectiveness of the criminal law lay at the core of the case. In contrast, the health care cases seem to raise more challenging issues related to allocation of resources and medical expertise. These cases might prompt the Court to be more cautious in finding a substantive violation of Article 2, since inter alia it might be difficult to assess the causal connections between the alleged inappropriate medical treatment and the harm sustained by the specific person. Perhaps for this reason, many of the cases in this area conclude only with a finding of a procedural violation (that is, failure to set up an independent judicial system so that the cause of death of patients in the care of the medical profession can be determined and those responsible held accountable).198 In this way and as already clarified in Section 6.B above, the Court avoids engagement with difficult questions of causation. The difficulties in establishing causal connections might also invite the Court to maintain the distinction between systemic and incidental failures. When confronted with information about deficiencies of a more systemic nature, the Court might more readily link the concrete case with these general shortcomings. 7. Conclusion No hard-edged legal tests apply to cases invoking positive obligations under the ECHR. This flexibility is similarly applied to the requirement for causation, that is, the linkage between the harm sustained by the individual applicant and state omissions. While certainty is not required that the interposition of a missing action would have prevented the harm, no general threshold has been articulated as to how likely it is that a protective measure would have averted the harm. The Court also merges issues of knowledge, reasonableness and causation in its assessment of state responsibility in the framework of positive obligations. Human rights law is thus far from rigid in the assessment of the linkage between state omissions and harm, an approach that can be understood in light of the objective of this body of law, namely assessing the responsibility of a collective (that is, the state). This assessment is underpinned by the assumption that the state is the entity tasked to ensure the rights of the individuals within its jurisdiction. As a consequence, the ECtHR’s approach wavers between effective protection of human rights, on the one hand, and not imposing an unreasonable burden on the state on the other. The establishment of causation is influenced by these considerations. Still, it would not be satisfactory to simply say that the standard of causation applied oscillates between effectiveness and reasonableness. Analytical rigour demands that we further scrutinize the role of causation in the context of positive obligations. This scrutiny shows that by assuming control over certain activities, the state places itself in proximate relationships with harm that might arise in relation to these activities even if this harm is not directly attributable to state agents. Control thus implies closer proximity and more demanding positive obligations. Paradoxically, in certain areas the absence of sufficient control by the state creates the basis for the finding that the state has failed to fulfill positive obligations. This paradox is perhaps only apparent since these areas are underpinned by the normative assumption that the state should assume control. This shapes the approach to causation by making it less stringent. Establishing causation between harm and state omission may be fraught with factual and epistemological uncertainty. In these circumstances, a conclusive determination that the nexus between state omission and harm is too attenuated or sufficiently solid to sustain a violation might be eschewed. Instead, we might rather ask whether the omission was contrary to the applicable domestic legal framework. In cases of non-compliance with this framework, the Court is more prepared to find that the omission has led to harm. Another avenue for avoiding issues of causality is by focusing on the process at national level leading to a decision that is allegedly contrary to states’ positive obligations. If this process is of sufficient quality, the finding of a violation is less likely. The assessment of the quality of the process also includes the availability of procedural guarantees. If these are incorporated at national level, the finding of a violation is also less likely. Finally, I have elaborated on three techniques that might limit the finding that the state is responsible. The standard of reasonableness, and the requirement that the risk of harm is ‘real and immediate’, might have such limiting functions. Although still with uncertain contours, the distinction between incidental and systemic failures might also be determinative. As a consequence, even if a failure is causative to harm, when the concrete case is representative of a mere incidental failure, no state responsibility might be found. ACKNOWLEDGMENTS The author would like to thank Matthew Williams Saul and Laurens Lavrysen for their comments on earlier drafts of this article. Responsibility for the views expressed remains the author's own. This article is part of a larger project on positive obligations in human rights law. Funding by the Ragnar Söderbergs Foundation is gratefully acknowledged. Footnotes 1 Costa, ‘The European Court of Human Rights: Consistency of its Case-Law and Positive Obligations’ (2008) 26 Netherlands Quarterly of Human Rights 449 at 453. 2 Stoyanova, Human Trafficking and Slavery Reconsidered. Conceptual Limits and States’ Positive Obligations in European Law (2017) at 319; McQuigg, ‘Domestic Violence as a Human Rights Issue: Rumor v. Italy’ (2016) 26 European Journal of International Law 1009. 3 Koch, Human Rights as Indivisible Rights: The Protection of Socio-economic Demands under the European Convention on Human Rights (2009); Mowbray, The Development of Positive Obligations under the European Convention on Human Rights by the European Court of Human Rights (2004) at 181. 4 Generally, authors have broken down their analysis depending on the specific subject areas covered in the case law: see Haijev, ‘The Evolution of Positive Obligations under the European Convention on Human Rights – by the European Court of Human Rights’ in Spielmann, Tsirli and Voyatzis (eds), The European Convention on Human Rights: A Living Instrument. Essays in Honour of Christos L. Rozakis (2011) 207. 5 The ECtHR itself has referred to two types of positive obligations: procedural and substantive: see Öneryildiz v Turkey Application No 48939/99, Merits and Just Satisfaction, 30 November 2004 [GC] at paras 89–96; Dröge, Positive Verpflichtungen der Staaten in der Europäischen Menschenrechtskonvention (2003); Starmer, ‘Positive Obligations under the Convention’ in Jowell and Cooper (eds), Understanding Human Rights Principles (2001) 139; Mowbray, supra n 3. 6 See O’Keeffe v Ireland Application No 35810/09, Merits and Just Satisfaction, 28 January 2014 [GC], where Judges Zupanicic, Gyulumyan, Kalaydjieva, De Gaetno and Wojtyczek in their Partly Dissenting Opinion (at para 7) observed how the scope of the positive obligations is affected by the principle of subsidiary. 7 Kratochvil, ‘The Inflation of the Margin of Appreciation by the European Court of Human Rights’ (2011) 29 Netherlands Quarterly of Human Rights 324; Lambert and Others v France Application No 46043/14, Merits and Just Satisfaction, 5 June 2015 [GC] at paras 144–148 (discussion of the margin of appreciation in the context of positive obligations). 8 Christoffersen, Fair Balance: Proportionality, Subsidiarity and Primarity in the European Convention on Human Rights (2009); Klatt and Meister, The Constitutional Structure of Proportionality (2012) at 86. 9 Plattform Arzte fur das Leben v Austria Application No 10126/82, Merits and Just Satisfaction, 21 June 1988 at para 31; Mowbray, supra n 3 at 221. 10 Hakimi, ‘State Bystander Responsibility’ (2012) 21 European Journal of International Law 341 at 349. Since the Court has justified positive obligations with the principle of effectiveness, Pieter van Dijk has observed that there are no clear-cut criteria for determining whether and when effectiveness will be achieved: see Dijk, ‘“Positive Obligations” Implied in the European Convention on Human Rights: Are the States Still the ‘Masters’ of the Convention?’ in Castermans-Holleman, van Hoof and Smith (eds), The Role of the Nation-State in the 21st Century. Human Rights, International Organizations and Foreign Policy. Essays in Honour of Peter Baehr (1998) 17 at 22. On the unpredictability of the ECtHR’s reasoning in relation to positive obligations, see also the Dissenting Opinion of Judge Kalaydjieva in Söderman v Sweden Application No 5786/08, Merits and Just Satisfaction, 12 November 2013 [GC]. See also Gerards and Brems, ‘Introduction’ in Brems and Gerards (eds), Shaping Rights in the ECHR. The Role of the European Court of Human Rights in Determining the Scope of Human Rights (2014) 1. 11 Xenos, The Positive Obligations of the State under the European Convention of Human Rights (2011) at 3; Than, ‘Positive Obligations under the European Convention on Human Rights: Towards the Human Rights of Victims and Vulnerable Witnesses’ [2003] Journal of Criminal Law 165 at 178. 12 Thielbörger, ‘Positive Obligations in the ECHR after the Stoicescu Case: A Concept in Search of Content?’ [2012] European Yearbook on Human Rights 259 at 261. 13 The Court has to navigate among different tensions and considerations: see Christoffersen, ‘Individual and Constitutional Justice: Can the Power Balance of Adjudication be Reversed?’ in Christoffersen and Madsen (eds), The European Court of Human Rights between Law and Politics (2011) 181. 14 Rights have a ‘dynamic aspect’ since they can generate new duties, which is fundamental for the understanding of their nature and function: see Raz, ‘On the Nature of Rights’ (1984) XCIII Mind 194 at 200. The strength of this approach is that it allows for flexibility for the emergence of new duties and the adaptation of the duties in light of the context and societal developments: see Report on the Right to Adequate Food as a Human Right submitted by Mr Asbjørn Eide, E/CN.4/Sub.2/1987/23, 7 July 1987 at para 47. On the dynamic aspect of rights, see also Gerards, ‘The Prism of Fundamental Rights’ (2008) 8 European Constitutional Law Review 173 at 178. 15 See also Hakimi, supra n 10 at 341, who has tried to identify such elements in the context of positive obligations under international law more generally. 16 McGrogan, ‘The Problem of Causality in International Human Rights Law’ (2016) 65 International and Comparative Law Quarterly 615 (the author argues that the UN monitoring system should focus on what can be known ‘rather than on abstract, aggregated quantitative measurements where causality cannot be plausibly attributed.’); Rigaux, ‘International Responsibility and the Principle of Causality’ in Ragazzi (ed.), International Responsibility Today. Essays in Memory of Oscar Schachter (2005) 81; Lavrysen, Human Rights in a Positive State (2016) at 137. 17 Ho and Rubin, ‘Credible Causal Interference for Empirical Legal Studies’ (2011) 7 Annual Review of Law and Social Science 17. 18 Fairgrieve, ‘Pushing the Boundaries of Public Authority Liability’ in Fairgrieve, Andenas and Bell (eds), Tort Liability of Public Authorities in Comparative Perspective (2002) 475 at 494; Booth and Squires, The Negligence Liability of Public Authorities (2006). 19 Husak, ‘Omissions, Causation and Liability’ (1980) 30 Philosophical Quarterly 318. 20 McGrath, ‘Causation by Omissions: A Dilemma’ (2005) 123 Philosophical Studies 125. 21 Steel, ‘Causation in Tort Law and Criminal Law: Unity and Divergence?’ in Dyson (ed.), Unravelling Tort and Crime (2014) 239; Fumerton and Kress, ‘Causation and the Law: Preemption, Lawful Sufficiency, and Causal Sufficiency’ (2001) 64 Law and Contemporary Problems 83. 22 Such an area is, for example, the extraterritorial application of positive obligations: see Ilascu and Others v Moldova and Russia Application No 48787/99, Merits and Just Satisfaction, 8 July 2004 [GC] at para 317. 23 L.C.B. v United Kingdom Application No 14/1997/798/1001, Merits and Just Satisfaction, 9 June 1998 at para 40. 24 Botta v Italy Application No 153/1996/772/973, Merits and Just Satisfaction, 24 February 1998 at paras 34–35. 25 Conforti, ‘Exploring the Strasbourg Case-Law: Reflections on the State Responsibility for the Breach of Positive Obligations’ in Fitzmaurice and Sarooshi (eds), Issues of State Responsibility before International Judicial Institutions (2004) 129 at 134–5 (emphasis added); Conforti, ‘Reflections on State Responsibility for the Breach of Positive Obligations: The Case-Law of the European Court of Human Rights’ (2003) 13 Italian Yearbook of International Law 3 at 3. 26 Rantsev v Cyprus and Russia Application No 25965/04, Merits and Just Satisfaction, 7 January 2010 at para 222; Mastromatteo v Italy, Application No 37703/97, Merits and Just Satisfaction, 24 October 2002 [GC] at para 76. See also Stoyanova, ‘Dancing on the Borders of Article 4: Human Trafficking and the European Court of Human Rights in the Rantsev Case’ (2012) 30 Netherlands Quarterly of Human Rights 163. 27 Xenos has also linked the elements of knowledge and causation: see Xenos, supra n 11 at 76. 28 O’Keeffe v Ireland, supra n 6 at para 144. 29 See Section 7.A where the obligation to take protective operational measures is addressed. 30 Brincat and Others v Malta Application No 60908/11, Merits and Just Satisfaction, 24 July 2014 at para 111. 31 The Court sometimes confuses the two contexts. See Section 7.B below. 32 K.U. v Finland Application No 2872/02, Merits and Just Satisfaction, 2 December 2008 at para 48. 33 In O’Keeffe v Ireland, supra n 6 at para 168, the GC established that the respondent state was aware in the 1970s of risks associated with sexual abuse of children by adults through, inter alia, ‘its prosecution of such crimes at a significant rate’. A similar approach was applied in Brincat and Others v Malta, supra n 30 at para 105, where the Court accepted that as early as 1987 laws were adopted to protect employees from asbestos and therefore since that date the state knew about the dangers associated with this substance. 34 Brincat and Others v Malta, supra n 30 at para 106. 35 Nencheva v Bulgaria Application No 48609/06, Merits and Just Satisfaction, 18 June 2013 at paras 121–2. 36 Cevriogli v Turkey Application No 69546/12, Merits and Just Satisfaction, 4 October 2016 at para 68. 37 Lavrysen, ‘Protection by the Law: The Positive Obligation to Develop a Legal Framework to Adequately Protect the ECHR Rights’ in Brems and Haeck (eds), Human Rights and Civil Rights in the 21st Century (2014) 69. 38 In some circumstances an initial assumption that the authorities had duties, and therefore ought to have known about risks of harm, might be warranted: see Premininy v Russia Application No 44973/04, 10 February 2011 at para 85 (prisoner beaten by other prisoners). 39 Van Colle v United Kingdom Application No 7678/09, Merits and Just Satisfaction, 13 November 2012 at para 96, where the Court determined that the harm was not foreseeable in the first place. See also Hiller v Austria, Application No 1967/14, Merits and Just Satisfaction, 22 November 2016 at para 53 (the psychiatric hospital staff could not have foreseen that a patient will escape and commit suicide). 40 Ugrekhelidze, ‘Causation: Reflection in the Mirror of the European Convention on Human Rights (A Sketch)’ in Calflisch, Callewaert, Liddell, Mahoney and Villeger (eds), Liber Amicorum Luzius Wildhaber Human Rights–Strasbourg Views (2007) 469 at 476. 41 See E. and Others v United Kingdom Application No 33218/96, Merits and Just Satisfaction, 26 November 2002 at para 99. 42 Salakhov and Islyamova v Ukraine Application No 28005/08, Merits and Just Satisfaction, 14 March 2013 at para 181 (‘Whether or not the authorities’ efforts could in principle have averted the fatal outcome in the present case is not decisive for this conclusion [failure to discharge a positive obligation]. What matters for the Court is whether they did everything reasonably possible in the circumstances, in good faith and in a timely manner, to try to save the first applicant’s life.’) 43 O’Keeffe v Ireland, supra n 6 at para 149; Opuz v Turkey Application No 33401/02, Merits and Just Satisfaction, 9 June 2009 at para 136; Premininy v Russia Application No 44973/04, Merits and Just Satisfaction, 10 February 2011 at para 84; Bljakaj and Others v Croatia Application No 74448/12, Merits and Just Satisfaction, 18 September 2014 at para 124 [emphasis added]. 44 Supra n 6 at para 166. See also E. and Others v United Kingdom supra n 41 at para 100. 45 Application No 52806/09 and 22703/10, Merits and Just Satisfaction, 5 December 2013 at paras 225 and 229. 46 Applications Nos 15339/02 et al., Merits and Just Satisfaction, 20 March 2008 at para 158. 47 Dodov v Bulgaria Application No 59548/00, Merits and Just Satisfaction, 17 January 2008 at para 97. 48 Draon v France Application No 1513/03, Merits and Just Satisfaction, 6 October 2005 [GC] at para 106. A clarification as to the application of the ‘direct and immediate link’ test in the context of Article 8 is due here. The Court seems to use the test to also determine whether the definitional threshold of Article 8 can be engaged in the first place. This can be related to the tendency of collapsing the definitional threshold enquiry, on the one hand, with the enquiry about the triggering and scope of positive obligations, on the other: see Gerards and Senden, ‘The Structure of Fundamental Rights and the European Court of Human Rights’ (2009) 7 International Journal of Constitutional Law 619. 49 Dubetska and Others v Ukraine Application No 30499/03, Merits and Just Satisfaction, 10 February 2011 at para 123. 50 Fadeyeva v Russia Application No 55723/00, Merits and Just Satisfaction, 9 June 2005 at para 92 [emphasis added]. 51 Stoyanovi v Bulgaria Application No 42980/04, Merits and Just Satisfaction, 9 November 2010 at para 61. 52 Giuliani and Gaggio v Italy Application No 23458/02, Merits and Just Satisfaction, 24 March 2011 [GC] at para 253. 53 Supra n 41 at para 100. 54 Application No 50385/99, Merits and Just Satisfaction, 20 December 2004 [GC] at para 63 [emphasis added]. 55 Application No 1967/14, Merits and Just Satisfaction, 22 November 2016 at para 53. 56 Another way in which causation matters in human rights law and in international law more generally, relates to remedies. In particular, once a violation of a right has been found, causation needs to be established between the violation and any harm for the purposes of awarding damages. See Article 41 ECHR. For an in-depth discussion on causation in the context of Article 41, see Kellner and Durant, ‘Causation’ in Fenyves, Karner, Koziol and Steinder (eds), Tort Law in the Jurisprudence of the European Court of Human Rights (2011) 449. This is an area of enquiry not pursued in this article. 57 Crawford, State Responsibility. The General Part (2013) at 113. 58 Draft Articles on Responsibility of States for International Wrongful Acts with Commentaries, Yearbook of International Law Commission, 2001, Vol II (Part Two) at 38, para 2. 59 Articles 4–7 ILC Draft Articles form the hard core of the doctrine of attribution since they deal with organs and agencies of state exercising sovereignty authority: see Crawford, supra n 57 at 115. 60 See Article 4(2) ILC Draft Articles; and Crawford, ibid. at 126. 61 Article 8 ILC Draft Articles. 62 See Article 7 ILC Draft Articles; and Armed Activities (DRC v Uganda) ICJ Reports 2005, 162 at 242. The difficulty here lay in distinguishing an official, though ultra vires, act from a purely private act: see Crawford, supra n 57 at 115. 63 Article 7 Draft Articles Commentary at para 18; and ILC Commentary, supra n 58 at 43, para 7. 64 I say ‘might’ because ‘[a]s a normative operation, attribution must be clearly distinguished from the characterization of conduct as internationally wrongful’: see ILC Commentary, supra n 58 at 39, para 4. 65 ILC Commentary, supra n 58 at 39 para 4. 66 ILC Commentary, supra n 58 at 47 para 1. 67 Contrast Military and Paramilitary Activities in against Nicaragua (Nicaragua v US) (Merits) [1986] ICJ Reports 14 at paras 109–115 (requiring effective control) with Case IT-94-1, Prosecutor v Tadic  (Appeal Judgement) IT-94-1-A, International Criminal Tribunal for the former Yugoslavia (ICTY), 15 July 1999 at paras 115-131 (adopting a standard of overall control). See also the confirmation of the effective control test in Application of the Convention on the Prevention and Punishment of the Crime of Genocide, ICJ Reports 2007, 43 at 207; and DRC v Uganda, supra n 62 at 226. 68 International law has accepted the distinction between primary and secondary rules of state responsibility. Primary rules are the substantive obligations in the various subject areas of international law. Secondary rules are those that elaborate on what it means for a state to be held responsible for violations of these duties. The rules on attribution belong to the latter. Positive obligations belong to the former: see Draft Articles on State Responsibility, supra n 58 at para 1. 69 It needs to be acknowledged, however, that on some occasions the ECtHR is not clear in its judgments as to why the respondent state can be held responsible: is it because harmful conduct is attributable to it or is it because it failed to fulfill its positive obligations: see Cyprus v Turkey Application No 25781/94, 10 May 2011 at paras 69–80. For discussion on the Court’s unclear logic, see Milanovic, ‘From Compromise to Principle: Clarifying the Concept of State “Jurisdiction” in Human Rights Treaties’ (2008) 8 Human Rights Law Review 411. 70 Ilascu and Others v Moldova and Russia supra n 22 at para 319. 71 Report of the ILC A/56/10 (2001) ‘Draft Articles’ at 44, (‘State’s authorities are strictly liable for the conduct of their subordinates; they are under a duty to impose their will and cannot shelter behind their inability to ensure that it is respected’). See also ibid. at para 319. 72 Report of the ILC A/56/10 (2001) Draft Articles at 102. 73 Evans, ‘State Responsibility and the European Convention on Human Rights: Role and Realm’ in Fitzmaurice and Sarooshi (eds), Issues of State Responsibility before International Judicial Institutions (2004) 139 at 157. 74 Medova v Russia Application No 25385/04, Merits and Just Satisfaction, 15 January 2009 at para 95 (disappearance case); Albekov and Others v Russia Application No 68216/01, Merits and Just Satisfaction, 9 October 2008 at paras 80–86 (it was not possible to establish who laid the mines which caused death; the Court did not have to decide on the issue since the respondent state was aware that mines were laid in the area and were under the positive obligation to protect the residents from the risks involved). 75 In the context of human rights law, there is a jurisdiction threshold which has to be passed so that the application of human rights treaties is triggered in the first place: see Milanovic, Extraterritorial Application of Human Rigts Treaties: Law, Principles, and Policy (2011). Control might be crucial for passing this threshold: see Lawson, ‘Life After Bankovic: On the Extraterritorial Application of the European Convention on Human Rights’ in Coomans and Kamminga (eds), Extraterritorial Application of Human Rights Treaties (2004) 83. 76 For this distinction, see Lawson, ‘Out of Control. State Responsibility and Human Rights: Will the ILC’s Definition of the “Act of State” Meet the Challenges of the 21st Century’ in Castermans-Holleman, Hoof and Smith (eds), supra n 10, 91 at 97. 77 Similar considerations have been made relevant in the context of claims against public authorities for damages at national level: see Brodie, ‘Compulsory Altruism and Public Authorities’ in Fairgrieve, Andenas and Bell (eds), Tort Liability of Public Authorities in Comparative Perspective (2002) 541 at 551 (‘I would suggest that where a public authority is concerned the court should, rather than seeking to identify pure omissions, look to see whether the authority in question has statutory responsibility to control, regulate, or supervise the relevant area of social or economic activity in the community’). 78 Sargyan v Azerbaijan Application No 40167/06, Merits, 16 June 2015 [GC] at paras 128–131; Besson, ‘The Bearers of Human Rights; Duties and Responsibilities for Human Rights: A Quite (R)evolution’ (2015) 32 Social Philosophy and Policy 244 at 253 (‘there is a general human rights’ positive duty for states to exercise jurisdiction and hence to incur human rights duties’). 79 Sargyan v Azerbaijan, ibid. at para 131. The following clarification is due here. In Sargyan v Azerbaijan, the issue under discussion was control for the purpose of establishing jurisdiction under Article 1 ECHR. However, the same logic can be extended in the context of positive obligations. On the use of the notion of ‘control’ for different purposes (establishing jurisdiction, attribution and positive obligations), see Besson, ‘Concurrent Responsibilities under the European Convention on Human Rights: The Concurrence of Human Rights Jurisdictions, Duties and Responsibilities’ in van Aaken and Motoc (eds), The European Convention on Human Rights and General International Law (forthcoming). 80 Finogenov and Others v Russia Application 18299/03 and 27311/03, 20 December 2011 at para 208. 81 See Article 2(2) of the ECHR which delineates limited circumstances when use of force might be justified. 82 McCann and Others v United Kingdom Application No 18984/91, Merits and Just Satisfaction, 27 September 1995 [GC] at para 194. 83 Supra n 52 at paras 253–262. 84 Application No 57950/00, Merits and Just Satisfaction, 24 February 2005 at para 188. 85 Supra n 54 at para 69. 86 Ibid. at para 70. 87 Supra n 80 at 209. 88 Ibid. at para 213 [emphasis added]. 89 Ibid. at para 214. 90 Ibid. at para 243. See also Mikayil Mammadov v Azerbaijan Application No 4762/05, Merits and Just Satisfaction, 17 December 2009 at paras 116–119, where the applicant complained that police officers were responsible for his wife’s death because upon their entry into the family’s dwelling she set herself on fire. The degree to which the state agents had control over the situation and the related degree to which they could have predicted the course of the events were essential in determining the scope of positive obligations under Article 2. 91 This paradox is further exaggerated if one takes into account that human rights are traditionally invoked to circumscribe the exercise of state control (states’ negative obligations to refrain), while positive human rights obligations might require more state control: see Reus-Smit, ‘On Rights and Institutions’ in Beits and Goodin (eds), Global Basic Rights (2011) 26 at 44. 92 Costello-Roberts v United Kingdom Application No 13134/87, Merits and Just Satisfaction, 25 March 1993; Dodov v Bulgaria, supra n 47; Storck v Germany Application No 61603/00, Merits and Just Satisfaction, 16 June 2005 at para 103; O’Keeffe v Ireland, supra n 6 at para 150. 93 Calvelli and Ciglio v Italy Application No 32967/96, Merits, 17 January 2002 [GC] at para 49; Vo v France Application No 53924/00, Merits, 8 July 2004 [GC] at para 89; Center for Legal Resources on behalf of Valentin Câmpeany v Romania Application No 47848/08, Merits and Just Satisfaction, 17 July 2014 [GC] at para 130; Lambert and Others v France Application No 46043/14, Merits and Just Satisfaction, 5 June 2015 [GC] at para 140 [emphasis added]. 94 One way will be to say that public functions are the ones that the state has historically performed. For a more in depth discussion, see Thomas, Public Rights, Private Relations (2015) at 41. 95 Supra n 6 at para 151. 96 Ibid. at para 162. This notion of ‘inherent obligation’ has not remained unchallenged. In their dissenting opinion, five judges challenged the assumption that there is some inherent risk of sexual abuse in the context of education (at para 15). 97 Ibid. at para 145 [emphasis added]. 98 Ibid. at para 157. 99 Application No 19986/06, Merits and Just Satisfaction, 10 April 2010 at para 35. 100 Storck v Germany, supra n 92 at para 103. 101 Dodov v Bulgaria Application No 59548/00, Merits and Just Satisfaction, 17 January 2008 at paras 84–86. 102 Application No 19202/03, Merits and Just Satisfaction, 24 April 2012 at para 63 (basic safety precautions were missing including not locking the door of the transformer located at a park for children). 103 Supra n 36 at para 57. 104 Ibid. at para 67. 105 Ibid. at para 69. 106 Reference was made to the various texts adopted by the Council of Europe in the field of environment and the industrial activities: Öneryildiz v Turkey, supra n 5 at paras 59–62 and 71. 107 Ibid. at para 71. Other industrial activities reviewed by the Court are nuclear testing (L.C.B. v United Kingdom Application No 14/1997/798/1001, Merits and Just Satisfaction, 9 June 1998), toxic emission from fertilizer factory (Guerra and Others v Italy Application No 14967/89, Merits and Just Satisfaction, 19 February 1998 [GC]) and exposure to asbestos at a workplace (Brincat and Others v Malta, supra n 30). 108 Vilnes and Others v Norway, supra n 45 at para 223. 109 Kolyadenko and Others v Russia Application No 17423/05, Merits and Just Satisfaction, 28 February 2012 at para 164. 110 Brincat and Others v Malta, supra n 30 at para 81 [emphasis added]. 111 Vilnes and Others v Norway, supra n 45 at para 223 (the Court ‘sees no need to consider in detail the degree of involvement of the respondent State in the hazardous activity in question, since the Convention obligation applies to “any activity, whether public or private”’). 112 Budayeva and Others v Russia, supra n 46 at para 174. The same distinction between man-made harm and natural harm is very clearly reflected in the Court's non-refoulement case law under Article 3 of the ECHR: see Stoyanova, ‘How Exceptional Must “Very Exceptional” Be? Non-Refoulement, Socio-Economic Deprivation and Paposhvili v Belgium’ (2017) 29 International Journal of Refugee Law 580. 113 Ibid. at para 137. 114 Ibid. at para 135 [emphasis added]. 115 Ibid. at paras 158-159. 116 Supra n 5 at para 134 [emphasis added]. 117 Ibid. at para 135. 118 Budayeva and Others v Russia, supra n 46 at para 173. 119 Ibid. at para 174. 120 Ibid. at paras 76–77. 121 I use the term ‘victim’ here in a general sense as a person who has sustained harm without prejudice to the determination whether the state can be held responsible under ECHR for this harm. 122 Keenan v United Kingdom Application No 27229/95, Merits and Just Satisfaction, 3 April 2001 at para 91. 123 Paul and Audrey Edwards v United Kingdom Application No 46477/99, Merits and Just Satisfaction, 14 March 2012 at paras 57–64 (a detainee was killed by another detainee while held in prison); Premininy v Russia Application No 44973/04, Merits and Just Satisfaction, 10 February 2011 at para 91 (a detainee was systematically beaten by other detainees). 124 Keenan v United Kingdom, supra n 122 at para 90 (the applicant argued that her son died from suicide due to the prison authorities failure to protect his life). 125 Florea v Romania Application No 37186/03, Merits and Just Satisfaction, 14 September 2010 (protection from passive smoking); Jashi v Georgia Application No 10799/06, Merits and Just Satisfaction, 8 January 2013 (provision of adequate care for detainee’s mental health). 126 Oette, ‘Austerity and the Limits of Policy-Induced Suffering: What Role for the Prohibition of Torture and Other Ill-Treatment?’ (2015) 15 Human Rights Law Review 669 at 681; O’Cinneide, ‘A Modest Proposal: Destitution, State Responsibility and the European Convention on Human Rights’ (2008) 5 European Human Rights Law Review 583. 127 Waiter v Poland Application No 42290/08, Admissibility, 15 May 2012 at paras 36-42 (access to life-saving drug); Budina v Russia Application No 45603/05, Admissibility, 18 June 2009. 128 Gerards, ‘The ECtHR’s Response to Fundamental Rights Issues related to Financial and Economic Difficulties: The Problem of Compartmentalization’ (2015) 13 Netherlands Quarterly of Human Rights 274. 129 Orchowski v Poland Application No 17885/04, Merits and Just Satisfaction, 22 October 2009 at para 153. 130 See also Nencheva v Bulgaria Application No 48609/06, Merits and Just Satisfaction, 18 June 2013 at para 119 (severely disabled children held in an institution died during the winter); Centre for Legal Resources on Behalf of Velentin Campeanu v Romania, supra n 93 at paras 134–144. This might also explain the Court’s lenient approach to causation in relation to victims of human trafficking held in immigration detention in the context of the application of Article 4 ECHR: see Stoyanova, ‘L.E. v. Greece: Human Trafficking and the Scope of States’ Positive Obligations under the ECHR’ (2016) 3 European Human Rights Law Review 290 at 299. 131 Denis Vasilyev v Russia Application No 32704/04, Merits and Just Satisfaction, 17 December 2009 at paras 115–116. 132 See Adam, R v Secretary of the State for the Home Department [2005] UKHL 66 (depriving asylum-seekers from social support when they are not allowed to engage in remunerated employment). 133 Keller v Russia Application No 26824/04, Merits and Just Satisfaction, 17 October 2013 at para 88. 134 Nencheva and Others v Bulgaria Application No 48609/06, Merits and Just Satisfaction, 18 June 2013 at paras 121–122. 135 Peroni and Timmer, ‘Vulnerable Groups: The Promise of an Emerging Concept in European Human Rights Convention Law’ (2013) 11 International Journal of Constitutional Law 1056; Ippolito and Sanchez (eds), Protecting Vulnerable Groups: The European Human Rights Framework (2015). 136 E. and Others v United Kingdom, supra n 41 at para 88. 137 M.S.S. v Belgium and Greece Application No 30696/09, Merits and Just Satisfaction, 21 January 2011 [GC] at para 232. 138 Placi v Italy Application No 48754/11, Merits and Just Satisfaction, 21 January 2014 at para 49. 139 Storck v Germany, supra n 92 at para 105. 140 Bevacqua and S. v Bulgaria Application No 71127/01, Merits and Just Satisfaction, 12 June 2008 at para 65. 141 D.H. and Others v Czech Republic Application No 57325/00, Merits and Just Satisfaction, 13 November 2007 [GC] at para 182. 142 Milanovic v Serbia Application No 44614/07, Merits and Just Satisfaction, 14 December 2010 at para 89. 143 One can distinguish this situation from ‘pre-existing vulnerabilities.’ For a discussion in the context of tort law, see Green, Causation in Negligence (2015) at 38. 144 I v Finland Application No 20511/03, Merits and Just Satisfaction, 17 July 2008 at para 44. 145 Cyprus v Turkey Application No 25781/94, Merits, 10 May 2001 [GC] at para 219; Waiter v Poland, supra n 127 at para 35. 146 Lopes de Sousa Fernandes v Portugal Application No 56080/13, Merits and Just Satisfaction, 15 December 2015 at para 114; Elena Cojocaru v Romania Application No 74114/12, Merits and Just Satisfaction, 22 March 2016 at para 111. 147 Ibid. at para 188. 148 Application No 55164/08, Merits and Just Satisfaction, 14 October 2010 at para 79. 149 Application No 46117/99, Merits and Just Satisfaction, 10 November 2004 (the national authorities did not comply with decisions by the national court ordering the closure of a mine); Giacomelli v Italy Application No 59909/00, Merits and Just Satisfaction, 2 November 2006 at para 93 (the state authorities did not comply with domestic legislation on environmental matters and failed to enforce judicial decisions); Kalender v Turkey Application No 4314/02, Merits and Just Satisfaction, 15 December 2009 at paras 43–47 (relatives of the applicants were killed in an accident at a railway station; the Court found a violation of Article 2 in its substantive aspect in view of the significant number and the seriousness of the breaches of the national safety regulations). 150 See generally, Lavrysen, supra n 37 at 82. 151 See, for example, Panaitescu v Romania Application No 30909/06, Merits and Just Satisfaction, 10 April 2012 at para 36. 152 Fadeyeva v Russia, supra n 50 at para 98 (after observing that the domestic legality is not a separate and conclusive test, the Court added that it is ‘rather one of many aspects which should be taken into account in assessing whether the State has struck a “fair balance” in accordance with Article 8(2)’). The problem with this approach is that the Court combines various factors, including non-compliance with the national legislation, into the general and very elusive standard of fair balance. This makes it impossible to objectively assess the role of and the weight attached to each individual factor, including non-compliance with domestic legislation. 153 Gerards and Senden, supra n 48 at 619. 154 Brems and Lavrysen, ‘Procedural Justice in Human Rights Adjudication: the European Court of Human Rights’ (2013) 35 Human Rights Quarterly 182; Brems, ‘Procedural Protection. An Examination of Procedural Safeguards Read into Substantive Convention Rights’ in Brems and Gerards (eds), Shaping Rights in the ECHR. The Role of the European Court of Human Rights in Determining the Scope of Human Rights (2014) 137; Gerards and Brems (eds), Procedural Review in European Fundamental Rights Cases (2017). 155 Öneryildiz v Turkey, supra n 5 at paras 89–90; Budayeva and Others v Russia, supra n 46 at para 132; Roche v United Kingdom Application No 32555/96, Merits and Just Satisfaction, 19 October 2005 [GC] at paras 161–162; Hatton and Others v United Kingdom Application No 36022/97, Merits and Just Satisfaction, 8 July 1997 [GC] at para 104. See also Steyn and Slarks, ‘Positive Obligation to Provide Access to Information under the European Convention on Human Rights’ (2012) 17 Judicial Review 308. 156 Roche v United Kingdom, ibid. at para 162. 157 Ibid. at para 161. 158 Hatton and Others v United Kingdom, supra n 155. 159 Ibid. at para 128. 160 Vilnes and Others v Norway, supra n 45 at paras 233–244. 161 Diving tables relate to the planning and the monitoring of the decompression. 162 Vilnes and Others v Norway, supra n 45 at para 233 [emphasis added]. 163 Ibid. at para 240. 164 Ibid. at para 238. 165 Ibid. at para 243. 166 Ibid. at para 244 [emphasis added]. 167 O’Keeffe v Ireland, supra n 6 at para 144. 168 See Hickman, ‘The Reasonableness Principle: Reassessing its Place in the Public Sphere’ (2004) 63 Cambridge Law Journal 166. 169 Mastromatteo v Italy, supra n 26 at para 74. 170 Ibid. at para 76 [emphasis added]. 171 For a useful outline of the interests involved, see the Partly Dissenting Opinion of Judge Bonello in Mastromatteo v Italy, supra n 26 at para 7. 172 Osman v United Kingdom Application No 23452/94, Merits and Just Satisfaction, 28 October 1998 [GC] at para 116. 173 Gerry, ‘Obligation to Prevent Crime and to Protect and Provide Redress to Victims of Crime’ in Colvin and Cooper (eds), Human Rights in the Investigation and Prosecution of Crime (2009) 423 at 432. 174 The interpretation advanced by McBride that the likelihood ‘has at least to be compelling’ is too stringent: see McBride, ‘Protecting Life: Positive Obligation to Help’ (1999) 24 European Law Review 43. 175 See Re W’s Application [2004] NIQB 67; Re Officer L [2007] UKHL 36 per Lord Carswell; Smith v Chief Constable of Sussex [2008] EWCA Civ 39; and Wright, ‘The Operational obligation under Article 2 of the European Convention on Human Rights and Challenges of Coherence–Views from English Supreme Court and Strasbourg’ (2016) 1 Journal of European Tort Law 58 at 67. 176 Ebert and Sijniensky, ‘Preventing Violation of the Right to Life in the European and the Inter-American Human Rights Systems: From Osman Test to a Coherent Doctrine on Risk Prevention?’ (2015) 15 Human Rights Law Review 343 at 359. 177 McBride, supra n 174. Xenos has also objected to the general applicability of the ‘real and immediate’ test. However, the substantiation of his objection is questionable. More specifically, he has observed that ‘the Osman test contains calculating parameters of “seriousness” and “immediacy” that confine the scope of positive obligations to mere reactive responses’ [emphasis added]. National authorities are, however, under the obligation to take protective operational measures upon risk of harm and before the harm has actually materialized. Certainly, once a case reaches the ECtHR, then an assessment of the national authorities’ reaction is made: see Xenos, supra n 11 at 112. 178 Opuz v Turkey, supra n 43 at paras 133-136; Rantsev v Cyprus and Russia, supra n 26 at para 286; Branko Tomasic and Others v Croatia Application No 46598/06, Merits and Just Satisfaction, 15 January 2009 at para 52. 179 See the Partly Dissenting Opinion of Judge Bonello in Mastromatteo v Italy, supra n 26. 180 In some judgments, the Court incorrectly applies the ‘real and immediate risk’ test. For example, in Öneryilidz v Turkey, supra n 5 at paras 100–102, the Court applied this test in its assessment as to whether Turkey was responsible for the deaths in the light of the substantive aspect of Article 2, that is, whether the national safety regulations were effective. The reference to the test in this context was, however, hard to defend. While the establishment that the national authorities knew about the dangers associated with the garbage collection point was necessary for the finding that state omissions were causative to harm (see Section 3 above), it is hard to understand why it needed to be also established that these dangers were immediate in the sense of the Osman test. In Öneryilidz v Turkey the notion of immediacy seems to be stretched to a breaking point (at para 100): ‘the risk of an explosion had clearly come into being long before it was highlighted in the report of 7 May 1991 and that, as the site continued to operate in the same conditions, that risk could only have increased during the period until it materialized on 28 April 1993.’ A period of two years is hardly what the Court had in mind when it established the ‘real and immediate risk’ test in Osman v United Kingdom. 181 There are judgments where the Court is very clear about the distinction between circumstances calling for protective operational measures and circumstances requiring general protection of the society: see Bljakaj and Others v Croatia Application No 74448/12, Merits and Just Satisfaction, 18 September 2014 at para 124; Stoyanovi v Bulgaria Application No 42980/04, Merits and Just Satisfaction, 9 November 2010 at paras 59 and 62; Mikhno v Ukrain Application No 32514/12, Merits and Just Satisfaction, 1 September 2016 at para 126. 182 For a useful outline, see Lavrysen, supra n 37. 183 See the Partly Concurring and Partly Dissenting Opinion of Judge Sajó in Banel v Lithuania Application No 14326/11, Merits and Just Satisfaction, 18 June 2013 (death of a boy after collapse of a roof). 184 The issue of acceptable level of risk has come to a head in the context of home births and state imposed regulations as to the conditions under which women can give birth: see the Dissenting Opinion of Judges Sajó, Karakas, Nicolaou, Laffranque and Kellder in Dubská and Krejzová v Czech Republic Applications Nos 28859/11 and 28473/12, Merits and Just Satisfaction, 15 November 2016 [GC] at para 29. 185 See Mastromatteo v Italy, supra n 26 at para 73, where it was observed that the percentage of crimes committed by prisoners on semi-custodial regime was very low. 186 Lopes de Sousa Fernandes v Portugal, supra n 146 at para 108; Lopes de Fernandes v Portugal, supra n 147 at paras 168 and 187; Elena Cojocaru v Romania, supra n 146 at para 100. Similar reasoning has been endorsed in other contexts too: see Stoyanovi v Bulgaria. supra n 181 at para 61. 187 As opposed to the Chamber in Lopes de Sousa Fernandes, supra n 146, which found a substantive violation of Article 2, the Grand Chamber, supra n 147, did not, in this way overruling the Chamber. For a more detailed analysis, see Lavrysen, ‘Medical Negligence after Lopes de Sousa Fernandes: A Blank Check to the Member States with Respect to the Substantive Right to Life?’ Strasbourg Observers Blog, 8 February 2018. 188 Lopes de Sousa Fernandes v Portugal, supra n 147 at paras 182, 190–192. In paras 191 and 192 of the judgment, the Grand Chamber explained the circumstances when ‘denial of immediate emergency care’ will transpire. These circumstances are framed as ‘very exceptional circumstances’. On the application of the ‘very exceptional circumstances’ standards, see Stoyanova, ‘How Exceptional Must “Very Exceptional” be? Non-Refoulement, Socio-Economic Deprivation and Paposhvili v Belgium’ (2017) 29 International Journal of Refugee Law 580. 189 Supra n 10 at para 50. 190 Ibid. at para 91 [emphasis added]. 191 The standards have not been applied in other cases either since Söderman v Sweden, ibid. 192 Ibid. at para 117. 193 O’Keeffe v Ireland, supra n 6 at para 166. 194 In Söderman v Sweden, supra n 10, the applicant invoked the ineffectiveness of not only the national criminal law remedies, but also of the civil law remedies. However, the civil law remedies were contingent on the criminal law remedies. 195 M.C. v Bulgaria Application No 39272/98, Merits and Just Satisfaction, 4 December 2003; Siliadin v France Application No 73316/01, Merits and Just Satisfaction, 26 July 2005. See generally Stoyanova, supra n 2 at 330; Lazarus, ‘Positive Obligations and Criminal Justice: Duties to Protect or Coerce’ in Zadner and Roberts (eds), Principles and Values in Criminal Law and Criminal Justice: Essays in Honour of Andrew Ashworth (2012) 136 at 150; Stoyanova, ‘Article 4 of the ECHR and the Obligation of Criminalising Slavery, Servitude, Forced Labour and Human Trafficking’ (2014) 3 Cambridge Journal of International and Comparative Law 407 at 414. 196 In the circumstances of unintentional killing, the Court has clarified that absence of a criminal law remedy might not be problematic: see Calvelli and Ciglio v Italy, supra n 93; and Vo v France, supra n 93. 197 Söderman v Sweden, supra n 10 at para 82. 198 Belenko v Russia Application No 25435/06, Merits and Just Satisfaction, 18 December 2014 at paras 84–85; Burzykowski v Poland Application No 11562/05, Merits and Just Satisfaction, 27 June 2006 at para 118; Dodov v Bulgaria, supra n 47 at para 98; Kudra v Croatia Application No 13904/07, Merits and Just Satisfaction, 18 December 2012 at paras 106–121; Bilbija and Blazevic v Croatia Application No 62870/13, Merits and Just Satisfaction, 12 January 2016 at para 119. © The Author(s) [2018]. Published by Oxford University Press. All rights reserved. For permissions, please email: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

Human Rights Law ReviewOxford University Press

Published: May 25, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off