Introduction
Twenty years have passed since the Human Rights Act 1998 (HRA) came into force. It is over 20 years since the decision of the European Court of Human Rights in Osman v United Kingdom.Footnote 1 Coinciding with this anniversary, the Supreme Court handed down two decisions in 2018 which raked up foundational issues about human rights, English tort law, and the Osman cause of action. Those decisions are DSD v Commissioner of the Metropolitan Police Footnote 2 and Robinson v Chief Constable of West Yorkshire.Footnote 3
The facts of Robinson and DSD are similar to Osman: all of them concern situations where the police have not prevented the claimant coming to grief at the hands of a third party.Footnote 4 That particular factual scenario is the focus of this paper. Claimants have sought remedies for it on the battlefields of tort law and human rights law. They have, generally, been more successful in human rights than in tort. The question is why. One of the unsolved mysteries of these last 20 years has been, ‘what is the relationship between tort and human rights law?’ I look at that question through the prism of this particular factual configuration: the state's failure to prevent harm from third parties. One of the things I shall be arguing is that this finer-grained approach can make more progress than some of the scholarship to date.
Whilst there is extensive literature on the functions of tort law, and of individual torts, the same service has not been performed for individual causes of action in human rights. Before that is done, little headway can be made with understanding the differences between human rights and tort. This paper explores the theoretical foundations of the Osman cause of action, but it is my hope that others will take up this project too.
The question of the relationship between tort and human rights is particularly acute with Osman because the ECtHR's formulation of the duty closely resembles the tort case of Dorset Yacht v Home Office:Footnote 5 where the authorities knew or ought to have known of a real and immediate risk of serious harm from a third party to an identified individual, they must take measures within the scope of their powers which, judged reasonably, might have been expected to avoid that risk.Footnote 6 Indeed, ECtHR judgments sometimes use the language of ‘negligence’ and a ‘duty of care under Article 2’.Footnote 7Osman looks like a modified negligence standard. But looks can be deceiving. The Supreme Court has now vehemently drawn a line between the HRA and tort. I examine what its portrayal of the law tells us about the old question on the relationship between tort and human rights. This paper has both a positive and negative project. I begin with the negative project, explaining why the approach of much literature to date is unconvincing. If the negative project persuades, then the contribution of this paper will be realised. The positive project is more ambitious and tentative. It will benefit from the input of other voices.
(a) The debate to date
There was vigorous debate, in the first 20 years of the HRA, about which taxonomy it fits into. Some scholars and judges placed the HRA within public law.Footnote 8 By contrast, an illustrious contingent classed it with tort law.Footnote 9 Yet others adopted an intermediary position between these two camps, arguing that it is a hybrid of tort/public law.Footnote 10 A fourth suggestion is that human rights form a standalone category by themselves.Footnote 11
Adopting a position on the question ‘what is the relationship between torts and human rights law?’ has certain consequences. If you think they are the same thing, the inquiry ends there. But if you think they are different, then there is a further question about whether they are compatible with one another. Even if you think they are different from one another, you may still think they are compatible – that tort law can adapt to meet human rights requirements. Thus, for some participants in this debate, it was a project in self-understanding, with an eye to which body of principle should be used to develop the law.Footnote 12 Their opponents saw this as a threat to the coherence of traditional bodies of law, and wanted to keep human rights out of tort.Footnote 13 By contrast, those attracted to the view that human rights are part of tort have a horror of the meagre damages in human rights cases.Footnote 14 Tort law's remedial doctrines are thought by them to form a laudable template. Their opponents worry about overburdening the public purse.Footnote 15 Tort is also thought by some to be the body of law that paradigmatically protects rights.Footnote 16 It follows for them that the HRA is axiomatically a tort statute. In turn, other interlocutors were concerned to promote the view that human rights are sui generis, so as to take that corpus seriously and retain as distinctive their considerable clout in legal-political terms.
My portrayal of this debate, which occurred over two decades with diverse motivations and nuanced positionings, is necessarily oversimplified. However, what can categorically be said about this scholarship is that it followed the courts’ methodology. The courts tend to identify features of doctrine – of procedural and remedial doctrine – that differ between tort and human rights law, and then draw conclusions about the significance of this. True to form, Lord Hughes in the recent case of DSD said the following features were significant. First, ‘the limitation period differs’. Secondly, ‘there is a more relaxed approach to causation in a Convention-based claim’. Thirdly, there is ‘a difference of approach to the calculation of compensation’.Footnote 17 The negative project undertaken in this paper is to show why focusing on these features is incomplete as a mode of analysis. That will be the work of the first part of the paper.
The second part does the positive project of theorising the Osman cause of action. There has been a burgeoning of useful scholarship in private law theorising individual causes of action. Yet we do not know much about human rights causes of action. Lord Mance in DSD was concerned by this, because ‘shaky foundations or rationale’Footnote 18 make it difficult to tell how far the cause of action extends. Theorising the cause of action can assist workaday courts, therefore. It also contributes to public debate. The Supreme Court's critique has not been confined to its judgments, but has spilled over into public consciousness. In the BBC Reith Lectures of 2019, Lord Sumption accused the ECtHR of ‘mission creep’ partly because it formulated positive duties which are not in the text of the Convention. A chorus of scholarly voices join him in criticism of the ‘open-ended’ scope of positive human rights duties,Footnote 19 which have ‘put the concept of positive obligations into disrepute’.Footnote 20 The ECtHR itself has decried any assistance, saying it is ‘not desirable, let alone necessary, to elaborate a general theory’.Footnote 21 Theoretical clarity may help us to move beyond this impasse. We will be in a better position to determine whether positive duties serve any valuable function, or whether they are instead troublesome transplants that undermine domestic legal categories.
Early pioneers of human rights scholarship, like StarmerFootnote 22 and Mowbray,Footnote 23 identified five kinds of positive duties which the Convention gives rise to.Footnote 24 Of these, the operational duty first identified in Osman is one. It is these sorts of duty that I refer to as ‘causes of action’ – ie actionable duties. Arguably more than five types of such duty now exist in the Court's mature case law, depending on how one individuates them. Our understanding of them is less advanced than it could be because of the method of analysis that commentators have previously used. Once we see the weaknesses of this method, the way is clear to theorise the causes of action. We begin by examining the factors that DSD identified as distinguishing tort and the HRA: remedy and limitation. We will see that scholars’ methodology inherits a preoccupation with these factors due to deeper differences over the correct way to conceptualise domestic legal categories. This intellectual baggage falls into perspective once the positive project in this paper is carried out, a project that puts the positive duty itself centre-stage.
Foregrounding the positive duty in this way reveals the significance of its key features. Human rights takes seriously the nature of the state-defendant in a way that tort law does not. So the second half of the paper suggests some political theories that may explain this international law tradition. By contrast with a Diceyan-inspired tort law, the state that owes its citizens no more than they owe one another is too minimalist for human rights law. State services are conceived of as entitlements rather than mere benefits under ECtHR doctrine. Further, the act/omission distinction, crucial to tort law as expounded in Robinson, matters less in human rights because the state is simultaneously connected with everyone in its jurisdiction. With that summary of the argument in mind, let us turn to the detail of this paper.
1. The negative project
(a) Remedy
Oft-cited differences between tort and the HRA include the availability, function, and quantum of damages. We look at each of these.
(i) Availability
Tortious liability gives rise to damages as of right,Footnote 25 whereas they are only available under the HRA if necessary for ‘just satisfaction’.Footnote 26 For some, like Duncan Fairgrieve,Footnote 27 this is a crucial classificatory distinction. Nolan finds this ‘particularly noteworthy’, and supportive of his view that the HRA belongs to the realm of public law.Footnote 28 It is also key for Cane, but he derives the opposite conclusion to Nolan. Cane puts the HRA within tort because he thinks monetary compensation is ‘unknown’Footnote 29 to administrative law. Juridical inequality between state and citizen is, according to Cane, what accounts for the difference between compensation and just satisfaction.Footnote 30 Any theory of tort law must accommodate litigation between parties who are not juridically equal, he says.
At stake in these debates are competing success conditions for theories of tort and public law. Arguments about the HRA map onto larger debates about the proper boundaries of the candidate taxonomies into which the HRA might fall, ie ‘private’ and ‘public’ law. These larger debates are not easily resolved, and will not tell us much about human rights while they remain contested.
But if the aim is to look at doctrine on the availability of damages, then there is more doctrine to look at than the language of ‘just satisfaction’. Even if the conclusions we could draw from the language of just satisfaction were safe, there is more doctrine which unsettles this conclusion – from Strasbourg. The approach of UK courts to damages can be contrasted with the remedial powers of the ECtHR. Any remedy or relief can be awarded under the HRA which is ‘just and appropriate’.Footnote 31 This is a considerably wider power than that available to the ECtHR even. The only remedy available to the ECtHR is damages.Footnote 32 Damages are awarded on an ‘equitable assessment’.Footnote 33
The reference to ‘just satisfaction’ appears both in the HRA and the Convention. In this jurisdiction, it is taken to refer to whether damages ought to be awarded in preference to another remedy, whereas the ECtHR has understood it to refer to the measure of damages. Just satisfaction has been interpreted by the ECtHR as a lesser measure than restitutio in integrum (restoration to original condition), and is awarded when restitutio in integrum is impossible.Footnote 34Restitutio in integrum is, of course, the usual measure of damages in tort.Footnote 35Osman-type situations are particularly interesting here, because they are hard cases for assertions that damages are unimportant. Damages are routinely sought; the breach is not ongoing (so injunctions are inappropriate) and the harm is undoable. It is overreaching, then, to conclude ‘monetary remedies are peripheral to the HRA’Footnote 36 or that ‘the concern will usually be to bring the infringement to an end’.Footnote 37
It might be said against me that the ECtHR's approach to damages is neither well-developed nor consistent. That is true. But this objection tends to confirm that remedial practice is too narrow a base upon which to build a theory of human rights law. Remedies are responses to something. Until we know what it is the remedy is responding to, it seems premature to base our conclusions on the type of remedy available.
(ii) Quantum
The same point can be made about quantum of damages. It all depends on whether one assumes that it is the same wrong being addressed by damages. If it is a different wrong, then measurement by different principles is not incongruous. (This does not foreclose anyone from arguing that, as a society, we are undervaluing human rights in the level of damages awarded.)
The literature to date has simply not given a detailed account of the wrong. An account of the wrong is not the same thing as determining what the aim of the remedy is (whether compensatory, vindicatory, etc), because that also may affect the quantum of damages awarded. We turn next to this.
(iii) Function
There is considerable agreement that remedies serve a different function under the HRA as compared with tort. In an early case, Lord Bingham denied the HRA was a tort statute on the grounds that its objects were ‘different and broader’.Footnote 38 Most recently DSD, echoing previous decisions,Footnote 39 confirmed the two causes of action give compensation on different bases.Footnote 40
On the other hand, there is little consensus about which particular functions tort and HRA remedies respectively pursue. In Smith, Lord Brown elaborated upon what he thought those functions were: ‘Where civil actions are designed essentially to compensate claimants for their losses, Convention claims are intended rather to uphold minimum human rights standards and to vindicate rights’.Footnote 41 Against this, Varuhas argues that there exist some vindicatory torts (such as those actionable per se) and that damages under the HRA should be modelled on them.Footnote 42 He would not restrict HRA damages to pursuing vindication, however, but advocates for remedial convergence more generally. For him, the two areas pursue common functions and protect similar interests. Whether or not his conclusions are true, Varuhas’ line of thought shows how important a fine-grained focus on the cause of action is. For if we tar all torts with the same compensatory brush, we would never notice that some torts might not be compensatory. Let us not presume all human rights causes of action are the same as one another either.
Steele agrees that the courts’ reference to a functional separation is ‘oversimplified and fails to capture the breadth of functions associated with the law of tort and its remedies’.Footnote 43 For Steele, it is not only in respect of torts actionable per se that tort law vindicates rights, and she correctly observes that analysis is complicated by inconsistent usage of the terms ‘compensation’ and ‘vindication’.Footnote 44
It will be clear by now that the disagreement on the function of damages is located within a larger debate about the function of tort law and the HRA. The larger debate is not one I can take on here, but I take on one strand of it in the next subsection. For now, I explain why a focus on function of remedy is not data-rich enough to determine by itself the relationship between torts and the HRA.
There are several weaknesses inherent in the strategy of focusing on remedy. First, money is fungible. Its fungibility gives it an incredibly fluid expressiveness. It carries many meanings. It is difficult to say that it stands for any one thing. Doubtless, from the law's perspective, this is part of its attractiveness as a remedy. However, it does not aid clarity of analysis. There is a paucity of explanatory doctrine; quantum-only trials in human rights law are not prevalent like they are in tort. Money varies its expressiveness along one scale. Its variable is quantum, and this does little to reduce ambiguity. How are we to determine that £100 is compensatory and £110 is super-compensatory? Is vindication super-compensatory or sub-compensatory? Equally, a refusal to grant monetary relief is ambiguous because it leaves us to infer from the absence of remedy.
Even if we grant that it is possible to elicit the function of remedy from its form, this is not dispositive of the taxonomical question. Two different areas of law may have some remedial objectives in common whilst still being taxonomically diverse. For example, in arguing for the convergence of tort and HRA remedies, Varuhas points out that the ECtHR aims at restitutio in integrum, like tort law.Footnote 45 Yet the same can be said of contract law. Indeed, damages for distress and inconvenience are available in contract,Footnote 46 just as they are from Strasbourg.Footnote 47 A complaint of breach of contract is, however, markedly different from a complaint of breach of human rights. It is important, therefore, to have an idea about what the complaint is founded on. Thus the second weakness of focusing on remedy is that it only tells you a limited amount about function, if it can tell you anything at all. There are other features one might reasonably want to know about before coming to a conclusion about the relationship between tort and the HRA.
The remedy tells us what our response should be, but not what required responding to in the first place; or why it required us to respond. Indeed, the remedy may be serving a totally different function to the norm the breach of which pre-existed it. The reason for granting a remedy need not be the same as the reason for categorising the act as unlawful in the first place. For example, a norm requiring employers to give employees a copy of their employment contract might stipulate a conventional sum to be awarded to any employee who did not receive such document and who successfully claims against his employer for another reason. The damages given to the employee may be by way of apology, or token recognition of a right, or for other reasons. However, the function of the prior norm is to promote the activity in question. It is not self-evident that the remedy pursues this function, particularly if the conventional sum is too low to be effective at deterrence. In fact, such a norm can be found in the Employment Rights Act 1996, section 1 and Employment Act 2002, section 38 where the sum is only two to four weeks’ pay. This is the third weakness of focusing on remedy.
Fourthly, an upshot of the multivalent nature of money is that remedies can serve more than one purpose simultaneously. Lord Scott proposed this in Ashley: ‘Although the principal aim of an award of compensatory damages is to compensate the claimant for loss suffered, there is no reason in principle why an award of compensatory damages should not also fulfil a vindicatory purpose’.Footnote 48 This is Steele's view: private law vindicates the right of the individual claimant, while the HRA vindicates the Convention right of the individual on behalf of the population in general.Footnote 49 This may well be true but I hesitate to endorse it. It is difficult to convincingly read it off remedy alone, due to the open-textured symbolism of money.
Fifth, not all wrongs have a judicial remedy. These manifest in what Bagshaw calls ‘sympathetic refusals’.Footnote 50 Some wrongs might escape our notice if the focus is on remedial outcome alone.
For these reasons, remedy cannot be conclusive of the relationship between torts and the HRA. I do not suggest remedies have no place in theorising, merely that they tell us little in isolation. The relationship between rights and remedies is complex, and analyses of rights complement analyses of remedies.Footnote 51 The more compelling approaches to the question ‘what is the relationship between tort and human rights law?’ have slotted remedies into a multifactorial review.Footnote 52 So I proceed below to consider the other factors Lord Hughes identified: limitation and causation. Before that, I touch on one final suggestion of how to distinguish torts and human rights which relates to remedy but in a different way – corrective versus distributive justice.
(iv) Corrective vs distributive justice?
Francois du Bois believes human rights obligations instantiate distributive justice and therefore stand ‘in tension with the basic structure of tort law’.Footnote 53 It is not that du Bois claims distributive justice plays no role in tort, but it is merely a constraining role on the doing of corrective justice. He considers that pursuing distributive justice is, by contrast, the very aim of human rights law.Footnote 54 Hanna Wilberg,Footnote 55 Hugh CollinsFootnote 56 and others have also associated the HRA with distributive justice.
Du Bois alights on two aspects of tort which, for him, instantiate corrective justice. The first is causation, which I deal with in a later part of the paper. The second is the acts/omissions distinction.Footnote 57 The court in Robinson was also much impressed by this distinction. The majority was tempted to rationalise all tort law in terms of commission rather than omission.Footnote 58 What is the significance of this for corrective justice, according to du Bois? The significance seems to be that there are more positive obligations in human rights law compared with tort. But that is not enough, it seems to me, to show tort is all about corrective justice. What is corrective justice? In du Bois’ view, it is the Aristotelian category of justice that rectifies injustices which occur between a doer and sufferer of harm.Footnote 59 On that definition, there seems to be no objection to injustices which are breaches of positive duties, rather than negative duties, though. I can see no objection to liability for omissions from the perspective of corrective justice itself.
Perhaps du Bois would respond that I need to enrich my conception of corrective justice. To do so, of course, risks getting into contentious debates about the true nature of corrective justice, but let us leave that aside. Du Bois says corrective justice resolves conflicts between normative equals, and takes the sphere of liberty of each one to be sacrosanct so that neither party can intrude on the other or be made to assist the other.Footnote 60Robinson also was tempted by this characterisation of tort law. Robinson emphasised Dicey's view that public authorities are generally subject to the same responsibilities as individuals.Footnote 61 The significance of this is not, it seems to me, that tort law carries out corrective justice therefore. The significance is that, looking through the eyes of tort law, there is no difference at all between Joe Bloggs and HM Prison Service, or between the man on the Clapham omnibus and Transport for London. This is a theme that will be developed below. Du Bois recognises this when he says, ‘Public authority liability cases [in tort law] could be understood as turning on distributive justice instead, but it is also possible to see them as simply pitching claimant against defendant in a purely bilateral relationship’.Footnote 62 If this is right, then questions involving public authorities can be seen through a lens of distributive justice or through a lens of corrective justice. They do not exclusively raise distributive questions in human rights law, nor do they exclusively raise corrective questions in tort law.
Indeed, the reason to reject these accounts is that they are too categorical about the spheres in which corrective and distributive justice operate. Notwithstanding du Bois's acknowledgement that distributive justice has a role in tort law, it does more than merely constrain findings of liability (as he would have it). Gardner points out that corrective justice must deal with ‘endogenous’ questions of distributive justice; one of them being how to distribute rights to corrective justice.Footnote 63 That question is precisely the one at play in these cases (Michael, Van Colle/Smith, Robinson, and DSD) when the courts decide whether a claimant has a cause of action in tort, the HRA, both, or neither.Footnote 64 This is a more plausible description of the role distributive justice plays in the Osman-type case. So we need closer focus on the cause of action. Briefly, beforehand, let us see if the limitation period sheds light on the question.
(b) Limitation
Even if we grant that procedure can indicate anything about substantive law, the limitation period can only be circumstantial evidence of underlying principle. It is usually used as a subsidiary argument to buttress others,Footnote 65 and the various positions taken on it are motivated by the same deeper issues about the proper role and scope of domestic taxonomies.
It is hard to swallow the thought that, had Parliament given the HRA the same limitation period as a tort, it would have fundamentally altered the nature of human rights. A cluster of weak arguments cannot add up to being stronger than the sum of their parts. The main pillar, remedies, is not capable of supporting the whole.
It is true Lord Hughes had a third string to his bow in DSD, and that was the ‘more relaxed approach to causation’. Nolan and du Bois discuss causation too. There is something to this, but we can better see what it is once we have examined the substantive cause of action. Let us move on to the positive project of this paper then.
2. The positive project – examining the Osman duty
This half of the paper is more tentative. My aim is as much to demonstrate the size of the gap in the literature as it is to plug it. Plugging it will take more than a sketch, which is all that space here affords. I hope this paper makes a start which others can build on or disagree with.
(a) The explanandum
We are concerned with a situation where a public authority has failed to stop something bad befalling a citizen at the hands of another private individual. Any explanation of the cause of action needs to explain why the defendant is always a public authority, why the victim is always a private person,Footnote 66 and what connects those two to the third-party perpetrator. It needs to explain at what point a human right is invoked, and what link the right has with the bad outcome. It needs to explain why the harm is not specified.
This latter point is enormously important, in my view. It is arguable the Osman duty, or something very like it, arises across most (if not all) Convention rights. A former President of the ECtHR, speaking extrajudicially, confirmed: ‘There is no a priori limit to the contexts in which a positive obligation may be found to arise’.Footnote 67 Case law suggests it arises under the following Convention Articles:
• Article 2 (right to life), as in Osman itself;
• Article 3 (right to be free from torture, and inhuman and degrading treatment and suffering);Footnote 68
• Article 4 (prohibition on slavery and forced labour);Footnote 69
• Article 5 (right to liberty and security);Footnote 70
• Article 6 (right to fair process);Footnote 71
• Article 8 (right to respect for private and family life);Footnote 72
• Article 9 (freedom of thought, conscience and religion);Footnote 73
• Article 10 (freedom of expression);Footnote 74 and
• Article 11 (freedom of association).Footnote 75
Any explanation of the Osman duty needs to explain the extraordinary breadth of its application. It protects a range of interests against imminent threats from third parties; interests like bodily integrity, free speech, and liberty. Thus when a peaceful protest is threatened by a counter-protest, the state has a duty to step in and separate the two groups such that the protesters can exercise their free speech rights. Already this is looking different from tort law, but nothing in my argument turns on that because I cannot here defend this breadth of ECtHR doctrine in detail. (Sceptical readers may turn to the case law in the footnotes.)
The Osman duty is broad not only in terms of protected interests but also in terms of defendants. Osman has been pleaded against the police, against hospitals,Footnote 76 schools,Footnote 77 prison authoritiesFootnote 78 and local authoritiesFootnote 79 (as carers of children or as social landlords). So it is misleading to think the category of defendant in HRA claims is narrow just because it is always a public authority. The category of claimant is not narrow either. The duty may be owed to the public at large,Footnote 80 not just to identified individuals (as in tort).Footnote 81 The factual scenarios encompassed by the duty are similarly broad. It has been applied to situations as diverse as environmental disasters,Footnote 82 hate crime,Footnote 83 escape of prisoners,Footnote 84 bullying,Footnote 85 and domestic violence.Footnote 86 To reiterate, the scope of this paper is confined to private individuals threatening other private individuals, rather than the application of Osman to natural disasters and the like, because that is where the similarity to tort law has vexed courts and commentators. If I have shown in the process that there is much more theorising human rights law to be done, so much the better.
True it is that both tort and the HRA deal in fundamental and non-material interests. From this survey of the Convention case law, however, it seems like human rights law recognises more of them than tort law does. In view of its breadth, I am going to refer to the Osman duty as the ‘positive duty of prevention’ (PDP). There is a risk that referring to it as ‘the Osman duty’ will give the impression that the explanandum is limited to Article 2 cases,Footnote 87 whereas its breadth is significant.Footnote 88
This extensive list of Convention Articles raises the question of what in the stranger-danger situation invokes a human right. Is it the case that every time a stranger causes you personal injury, your human rights have been infringed? If not, what in the Osman formulation picks out those conditions? Are the consequences (personal injury, inhibited speech, invaded privacy…) the thing that engage your human rights? I take up these questions below.
I examine them by asking what a claimant would plead, in natural language. This is because, in my view, any explanation of the cause of action should be explicable to a layperson. If it turns out that human rights can only be understood and invoked by highly sophisticated persons, then we have failed to capture something vital: their universality and humanness. That said, let's proceed to the analysis itself.
(b) The nature of the act-complained-of
I adopt the view that breaching a human right is a ‘wrong’ in Gardner's sense of being a breach of duty.Footnote 89 That does not yet distinguish it from tort. So what sort of wrongs are they? The judiciary have some suggestions. In Attorney-General v Ramanoop, Lord Nicholls said, ‘the fact the right violated was a constitutional right adds an extra dimension to the wrong’.Footnote 90 A similar metaphor appeared in Michael but accompanied by a very different attitude. Lord Toulson was averse to ‘gold-plating’Footnote 91 the claimant's Convention right by adding a negligence cause of action. Both Lord Nicholls and Lord Toulson conjure the idea of a cause of action with a geometric physicality that can have a layer/dimension added by a concurrent cause of action. The difference is that Lord Nicholls conceives of the basic wrong as being in tort whereas Lord Toulson conceives of the basic wrong as being under the Convention. I argue that the difference is not merely an ‘extra dimension’ or metallic quantum sheen – ie the HRA is not simply ‘tort-plus’ per Lord Nicholls or ‘tort-minus’ per Lord Toulson. It is not slightly more wrong or slightly less wrong. Instead, the fundamental wrong being alleged in PDP cases is a distinct wrong.
In saying it is a distinct wrong I mean three things. First, it is a distinctive wrong. I argue that all PDP cases share an essential complaint. Secondly, it is distinct (separate) from any wrong committed by the third party who committed the assault. That is to say, liability for breach of a PDP is not vicarious or secondary liability for the third-party perpetrator. Thirdly, in saying that it is a distinctive kind of allegation, the argument is that it has elements distinct (different) from negligence as that was understood in Robinson and Michael. (Doubtless these decisions can be critiqued, but my theorising must fit the law as it is stated to be.) This third line of argument will be developed in the two sections following. The first two lines of argument are expanded on here.
First, then, all PDP cases have an essential complaint in common. The accusation is not that ‘you hit me’ – as it will be in many tort cases – but rather that ‘you did not stop him from hitting me’. The difference between these two allegations goes to the fundamental nature of the wrong. To see this, one can substitute the verb ‘hit’ in both accusations above, in order to make it a different act-complained-of. In the first sentence, it significantly changes the character of the allegation when we substitute the verb ‘hit’. To say ‘you killed me’ is a very different description of what occurred, and why it was wrong. Compare this again to ‘you detained me’ or ‘you slandered me’ or ‘you lied to me’.Footnote 92 In these examples, addressed to the third-party, changing the description of the allegation fundamentally changes the nature of the wrong.
By contrast, it matters less what verb-act you say the state failed to prevent from happening. The description of omission remains constant across the different iterations of the second type of accusation, viz ‘you did not stop him from ⱷ-ing me’ (where ⱷ can be substituted with ‘hit’, ‘kill’, etc, without changing the structure of the accusation). The third party's behaviour is not the focus of the description. Instead, attention is directed at the failure to stop it. When the allegation is addressed to the third party, the commission of each verb-act would require a different explanation of what was wrong about it. Whereas, with the allegation addressed to the state, how the wrong is committed and what makes it a wrong is much less dependent on the specific implications of the verb-act represented by ⱷ.
Thus the focus of this way of constructing the complaint is not so much that a bad thing happened, nor that some stranger was the perpetrator. Liability under a PDP is not secondary or vicarious liability for the perpetrator. It is the state's act/omission that breaches the human right; it is not the third party who breaches the right. Whilst it might be objected that human rights have ‘horizontal effect’ and apply to individuals,Footnote 93 recall that my explanandum is the PDP. Only public authorities are subject to it.
We can see this clearly when we examine the reasoning by which the ECtHR has defended positive duties. It relies on principles of state responsibility and attribution developed in international law.Footnote 94 To take an example, in Appleby v UK,Footnote 95 it was held the government did not bear ‘direct responsibility’ because the private company which restricted the applicant's freedom of speech could not be construed as a state agent. Rather, ‘the issue to be determined is whether the government have failed in any positive obligation’.Footnote 96 This is a clear statement that liability under the PDP does not rely on attributing to the state the culpability of the third party, but rather it is based on the state's own failure in a positive obligation.
Yet there does seem to be something parasitic about the liability which requires explanation. To ask whether liability is parasitic is to ask what the relevance of the outcome is, the bad consequence. It is to ask where the Convention Articles, and the interests they protect, fit in. The relevance of the bad outcome is that its occurrence is indicative that the state did not protect a right the claimant had. The fact of a breach is not retrospectively determined; rather, the outcome highlights that a breach may have occurred.Footnote 97 A claimant must be a ‘victim’ within the meaning of the HRA, s 7 in order to sue. It is not the stranger who breaches the claimant's human rights by harming her; it is the state that does by its failure to avert the threat.Footnote 98 The Convention does not create a freestanding right to (eg) bodily integrity, exigible against the state as well as the perpetrator of the harm. Thus a fuller expression of the essential allegation in a PDP case is: ‘By your failure to uphold my Convention right, another wrong was done to me by someone else’.
Faced with an accusation of ‘you did not stop him from ⱷ-ing me’, there are two obvious responses. Both are crucial constituents of the distinct type of wrong that breach of a PDP instantiates. The first response is: ‘how could I stop him?’. The second is ‘why should I stop him?’ Along with the distinctive form of the allegation, these responses reveal the other key features of the wrong.
As to the first question ‘how could I stop him?’, the accuser's reply would broadly be: ‘by an act of governance’.
As to the second question, ‘why should I stop him?’, the accuser's reply would be: ‘because of your responsibility toward me and for him’.
So the form of a PDP allegation invites two retorts, the answers to which reveal two essential components of this kind of wrong. The first component is a failure of governance. This is the act or omission complained of. The act-complained-of must manifest disregard for a protected aspect of human existence (ie a fundamental right or interest protected by a Convention Article). The conduct of the perpetrator poses a threat to a protected aspect of the complainant's existence; it is only of concern in the allegation to that extent. As we saw above, in the distinctive form of the allegation ‘you did not stop him from ⱷ-ing…’, the activity represented by ⱷ-ing matters much less than where the allegation is addressed to the third party.
The reason the ‘ⱷ-ing’ does not matter as much is that there are myriad and ingenious ways to pose threats to protected aspects of people's existence. Some of these may be torts or crimes in themselves, but they need not be. The criminal/civil liability of the perpetrator is a matter of domestic law, although there may be a separate human rights duty on the state to adopt a legal framework that prohibits certain harmful activities.Footnote 99 Unlike domestic law, human rights law is primarily concerned with the liability of states. A number of commentators have warned against assuming these two bases of liability are coextensive.Footnote 100
The thing that failed to be done is something which should be done by the state, which is why I described it as an act of governance. To give some simplified examples from cases: in Robinson, the alleged defective act of governance was the planning/execution of an arrest; in Michael and Sarjantson v Chief Constable of Humberside it was the response of the emergency services; in Thomson v Scottish Ministers, it was the decision to release a prisoner; and in Bedford v Bedfordshire County Council, it was the supervision of a private care home by the local authority fulfilling its statutory duties under the Children Act 1989.
For the Supreme Court in Michael, this was the explanation of why tort law sometimes subjects public authorities to the same standards as private citizens. The majority noted that there are some situations, usually involving the standard to be expected of professionals (eg NHS doctors, education, psychology),Footnote 101 where the duty of care owed is no different just because the defendant is a public authority. The Supreme Court said these situations were not ‘peculiarly governmental’Footnote 102 in nature. Assuming this is correct (as I must if I am not to opportunistically reject doctrine that does not fit my theory), it is beyond the confines of this paper to give a worked-out theory of which acts are peculiarly governmental, a topic on which there is a lot of literature.Footnote 103
However, the question of when to attribute acts to the state, or which acts have a public character, is one that human rights law has already faced.Footnote 104 In the modern state, when government services are frequently outsourced, both the domestic courtsFootnote 105 and ECtHRFootnote 106 have had to define the entities which count as ‘public authorities’ for the purpose of exercising particular governmental functions. Thus there is implicit recognition, in both the tort case of Michael and in cases adjudicating which entities exercise public functions under the HRA, that the concern is with the duties of the state. Unlike in tort, there is no assumption in HRA law that they are coextensive with the duties of the individual. Thus, the thing that was not done – the omission – ought to have been done by the state. An act of governance was called for.Footnote 107
Let's draw all this together in the intuitive form of accusation this section started with. The facts are taken from a case in the Norwegian Supreme CourtFootnote 108 but could equally be from a number of ECtHR casesFootnote 109 because they are unfortunately all too common. The allegation is:
– You did not stop him stalking me. Being stalked poses a threat to a protected aspect of my existence, ie my private and family life and psychological integrity insofar as that is dependent on my and my children's domestic security [which is an interest represented by Article 8]. As a consequence of your failure, I suffered mental distress.
– How should I have stopped him?
– By prosecuting his breaches of the restraining order [which is an act of governance].
– Why should I have stopped him?Footnote 110
– Because of your responsibility for him and toward me [ie you had a duty].
We have not yet discussed the second component of the cause of action, grounding the state's duty in responsibility for and toward diverse individuals. That comes next.
(i) Responsibility
Tort law has a characteristic discourse.Footnote 111 That discourse is very uncomfortable with the complaint ‘you had responsibility over him and toward me’. It is uncomfortable with both elements of that simple claim. Tort law's discomfort with responsibility toward the claimant manifests in the ‘conferral of benefit’ doctrine. Its discomfort with responsibility for the third party manifests in what has variously been called the ‘acts/omissions’ or ‘positive/negative duty’ distinction, which Robinson made much of.Footnote 112 Correspondingly, the PDP relies on a very different political theory that underlies human rights discourse. Let us expand on these points.
Taking responsibility-toward-the-claimant first, Lord Hoffmann has described the Osman case as one where the complaint was a failure to receive a benefit from the state.Footnote 113DSD referred to Lord Hoffmann's difficulty imagining a case ‘in which a common law duty could be founded simply upon the failure (however irrational) to provide some benefit which a public authority has a public law duty to provide’.Footnote 114 In a claim for breach of homelessness legislation, he articulated the following principle: the failure to provide a public service does not entitle the citizen to damages because the citizen is no worse off than if the service was never provided.Footnote 115 This is a crucial difference between tort and HRA law.
What tort regards as largesse, human rights regards as entitlement. Du Bois premises human rights law ‘on the notion that membership of a political community entitles individuals to certain services by the state’.Footnote 116 Shue elaborates on how this applies to the PDP kinds of cases under consideration here: ‘A demand for physical security is not normally a demand simply to be left alone, but a demand to be protected against harm. It is a demand for positive action’.Footnote 117 This is not just a demand, but an entitlement. Alexy explains: ‘Every right to a positive action on the part of the state is an entitlement. Rights to positive action impose upon the state… the purposes to be pursued’.Footnote 118
The minority in Michael argued, along Alexy's lines, that the state had failed in its policing purposes, and its autonomy was unlike the freedom of an individual. This kind of argument only removes a barrier to recognising liability, of course. Lord Hoffmann would still want to know why it sounds in damages paid to an individual. That is the sticking point. Yet negligence law does have a conception of responsibility-towards-the-claimant; but it's more local. Tort's narrower concept of responsibility-towards-the-claimant explains the requirement for an identified victim in tort law by contrast with the PDP, which may be owed to the public at large.Footnote 119 It also explains why the majority in Michael required an assumption of responsibility towards Ms Michael, namely because there was no general pre-existing tort responsibility.
Michael linked the doctrine that there is no tort entitlement to benefits with the second element of the responsibility that distinguishes the PDP, responsibility-over-the-third-party. Lord Toulson said:
It does not follow from the setting up of a protective system from public resources that if it fails to achieve its purpose, through organisational defects or fault on the part of an individual, the public at large should bear the additional burden of compensating a victim for harm caused by the actions of a third party for whose behaviour the state is not responsible. To impose such a burden would be contrary to the ordinary principles of the common law.Footnote 120
It is on this responsibility-over-the-third-party that we now focus our attention.
Du Bois locates the difference in conceptions about the state's role. For him, the prevalence of positive obligations under the HRA reflects ‘an understanding of the state as bearing special responsibilities in respect of those over whom it exercises authority’.Footnote 121 The House of Lords explicitly linked the act-omission distinction to responsibility: ‘The real issue in all these cases is whether the state is properly to be regarded as responsible for the harm inflicted (or threatened) upon the victim’.Footnote 122 Tort law's position on this, as described in Robinson, is that ‘public authorities, like private individuals and bodies, generally owe no duty of care towards individuals to prevent the occurrence of harm’.Footnote 123 The rationale usually given for tort law's chariness to impose positive duties to rescue is the value of an individual's freedom of action.Footnote 124 The dissent in Michael emphasised this cannot apply to the state in the same way:
Th[e police] should not be exempted from liability on the general common law ground that members of the public are not required to protect others from third party harm; such protection of autonomy for individuals is not appropriate for members of a force whose duty it is to provide precisely the type of protection from the harm that befell Ms Michael.Footnote 125
Like responsibility-towards-the-claimant, tort has a more local conception of responsibility-over-the-third-party than human rights law. The more local conception is seen in tort law's willingness to impose a duty when the third party is already under the control of the state (eg Dorset Yacht itself).
What could explain the understanding of state responsibility that du Bois attributes to the HRA? The ECtHR is cursory in its explanations. Mowbray observes that one common explanation is that positive obligations are required to make human rights ‘practical and effective’ rather than ‘theoretical and illusory’.Footnote 126 Quite apart from the fact that this pronouncement assumes its own truth,Footnote 127 the justification applies to all positive duties and cannot assist us with the PDP specifically.
More promising are references to Article 1, which puts the state under a duty to secure to everyone within the jurisdiction the rights and freedoms in the Convention. The ECtHR sometimes rationalises positive duties on this basis.Footnote 128 Besson elaborates that ‘there is a general human rights positive duty for states to exercise jurisdiction and hence to incur positive duties’.Footnote 129 Tort law's conferral-of-benefit doctrine would not sit well with Article 1.Footnote 130 Further, if Article 1 connects the state with everyone in the jurisdiction simultaneously,Footnote 131 the distinction between acts and omissions is less important than it appears to be in tort law. Indeed, the ECtHR frequently asserts that it does not matter whether a case is analysed in terms of positive or negative duties.Footnote 132
Article 1 cannot by itself get us all the way to the PDP, and more will have to be said specifically about the PDP below, but this point is worth dwelling on. It is worth dwelling on because, as a matter of public international law, there has been a shift toward recognising ‘R2P’ – ie responsibility to protect people from mass atrocities even outside the state's own jurisdiction. Article 1 ECHR is, of course, limited to where the state exercises its own jurisdiction.Footnote 133 The point here is not to digress into discussion of R2P but to note the common premise that R2P shares with the PDP. As a matter of customary international law, no state denies its duty to protect its own population from mass atrocities.Footnote 134 It is from this premise that discussions of protecting external populations proceed. Mass atrocities are presumably just egregious and large-scale breaches of certain Convention Articles.Footnote 135 (Indeed, some theorists go as far as to identify all human rights breaches – not just atrocities – as those where a state is susceptible to outside intervention should it fail to properly respond.Footnote 136) The interesting part is that tort law does not dissent from this premise wholeheartedly. In Robinson, the majority was content to say positive duties were owed to the public at large, but the point of departure was that they were not owed to individuals (save for where identified exceptions apply).Footnote 137 Tort law resists the idea that the general duty is an aggregate of the particular; it resists the move from the public at large as beneficiary to individual beneficiaries of duty. Human rights law does not.
With this tradition of international law in mind, let us consider which political theories help to ground the PDP specifically. There are a few candidate theories I will sketch here, but they are proposals rather than full arguments. Some candidate theories converge around the moral principle that with great power comes great responsibility. One possible way to get to the PDP is to argue that this principle places state officials under obligations ‘simply by virtue of their capacity to act’.Footnote 138 Certain political philosophers tie this to the social contract itself. Amanda Greene puts it this highly. She says, ‘Virtually every system of political rule advances the claim that it (at least) provides for the basic security of those it addresses as subjects’.Footnote 139 This is the claim of minimally-competent governance. She clarifies that basic security means minimal order and protection from private violence. It is a necessary condition for legitimate government that consent to be governed (from which the state derives sovereignty) is provided on the basis of the claim to minimally-competent governance. Following Bernard Williams, she says this is what separates political order from slavery or war.Footnote 140 The result is that the state which breaches the PDP breaches the social contract. It would seem the ECtHR puts the duty slightly higher than Greene, because the PDP arises under a range of Convention Articles and is not limited to minimal order and freedom from violence.
You could alternatively come to the PDP by a different route. This is the theory of structural violence.Footnote 141 Structural violence is often contrasted with direct violence, and direct violence has traditionally been the focus of law. Structural violence is ‘the aspect of violence of the differentiation of the social system… and the state is actively involved in retaining the status quo of that differentiation’.Footnote 142 A concrete example from a case will illustrate. In Bedford, a boy under local authority care assaulted another boy. The court noted the assailant's history of neglect, abuse, drug use and criminal record. A theory of structural violence would hold: ‘if there is an already violent reason for the violence of a person, to prohibit only the latter's violence in effect legitimates the former violence’.Footnote 143 The direct private violence of the individual then
legitimates the state because of the reactive violence it exercises against such private violence. The very notion of the state's ‘monopoly of violence’… serves as a violent construct to legitimate the fact that the state sanctions private violence that is the result of structural violence furthered, in part, by the state.Footnote 144
All this is to say that earlier failings or structural inequality on the part of the state may result in direct violence perpetrated further down the line. This is one possible way to ground the state's responsibility for the third party in PDP cases. The concept of structural violence contrasts strongly with tort law's traditional notions of individual responsibility, which (from the point of view of the structural violence theory) is blind to anything but direct private violence. This does not mean structural violence undermines or is inconsistent with individual responsibility,Footnote 145 merely that it is a less ‘local’ concept of responsibility-for than tort law traditionally has. The concept of structural violence is already found within areas of law that are influenced by human rights discourse, such as international law,Footnote 146 transitional justice,Footnote 147 and economic, social and cultural rights.Footnote 148 Some support for this idea can be found in ECtHR doctrine on the investigative duty at issue in DSD, but it has not been said to ground Osman. The ECtHR said that ‘what is at stake here is nothing less than public confidence in the state's monopoly on the use of force’.Footnote 149 This rationale for the PDP is limited, though, as it only explains the case of a dispossessed assailant, as in Bedford.
These kinds of theories only partially explain the scope of the PDP, but they illustrate why the conferral-of-benefits doctrine does not fit in human rights law. The state that owes its citizens no more than they owe each other is too minimal a conception of the state for human rights law. We can arrive at the PDP through the notion of structural violence, through great-power responsibility, breach of the social contract, through a combination of these, and probably via a number of other routes. I do not mean to advocate for any particular view but to show the array. The main thing to take from this discussion is how alien to English tort law these arguments are. Contained within duty are assumptions about responsibility. Tort law's assumptions tend to be more local than the HRA's. The underlying political theory of the HRA is more consonant with a role for the state as a guarantor of rights.
Lord Hoffmann's argument that the citizen is no worse off when a benefit is not provided bears on the act/omissions distinction and tort law discourse on causation. The decision on causation which allowed the claimant to succeed in Robinson was of this sort. The Supreme Court held Mrs Robinson was injured as ‘a result of being exposed to the very danger which the officers had a duty to protect her’.Footnote 150 In the Scottish case of Thomson, the court held that releasing a prisoner was not the same as creating a danger by releasing a dangerous thing because of ‘the very existence of third party action as the proximate cause of the injury’.Footnote 151 Thus, responsibility-for-the-third-party is also reflected in tort law's understanding of causation which is considered next.
(c) Causation
We circle back, finally, to suggestions that causation differs in human rights. The difficulty in PDP cases, from the point of view of tort law, is ‘there is always going to be someone more obviously to blame’Footnote 152 than the state. That someone is the perpetrator. However, what I hope to have shown above in my conceptualisation of the wrong is that the PDP complaint is of a different sort to the one directed at the third party. There are two wrongs, and responsibility for them is not a zero-sum game. The perpetrator is only ‘more obviously’ to blame if the underlying assumptions of tort law are drafted into actions under the HRA.
Tort law's unwillingness to impose responsibility-for-the-third-party makes it seem ‘obvious’ that the third party is to blame. But all that changes once there is a duty to guard against particular harms.Footnote 153 As the House of Lords recognised, ‘It would make nonsense of the existence of such a duty if the law were to hold that the occurrence of the very act which ought to have been prevented negatived causal connection between the breach of duty and the loss’.Footnote 154 We can now see how focusing on duty illuminates the causation argument.
The effect of the alternative approach to causation is clear from Sarjantson and the dissent in Michael. In Sarjantson, a HRA case, the defendant police argued their delayed response made no difference, because they would have arrived after the violence had started even if they responded promptly. The Court of Appeal held this would affect quantum but not liability. It is instructive to compare the majority and minority judgments in Michael too. The matters that the majority thought went to the question of duty (ie whether responsibility was assumed) were, for the minority, more relevant to causation. Lord Toulson, for the majority, said: ‘The only assurance which the call handler gave to Ms Michael was that she would pass on the call to the South Wales Police… She told Ms Michael that they would want to call her back and asked her to keep her phone free, but this did not amount to advising or instructing her to remain in her house…’Footnote 155 Lord Kerr countered, ‘While the police are not responsible for the actions of her murderer, if the allegations made against them are established, police played a direct, causative role in her death’.Footnote 156 Lord Kerr recognised that calls to the police impact people's actions, and it is likely that Ms Michael remained at home expecting help.
Lady Hale, also dissenting, made explicit the role of the state:
A person faced with the threat of violence is permitted by law to take reasonable measures of self-protection, but beyond that her only option is to inform the police. In essence, other than reasonably protecting herself, the law obliges her to entrust her physical safety in the police.Footnote 157
The state has a monopoly on the legitimate use of force, as the ECtHR recognised.Footnote 158 There is limited scope for individuals, particularly vulnerable ones, to protect themselves physically and they therefore rely on the state. Lady Hale's judgment has echoes of great-power-responsibility theory.
These competing conceptions of causation trace back to different theories of the state. For example, the theory of structural violence helps to explain the state's role in creating the conditions for seemingly random acts of violence. So although it is said the HRA has a ‘looser’Footnote 159 or ‘more relaxed’Footnote 160 approach to causation, it would be better to say it has a different approach to causation.Footnote 161 Its approach reflects its foundational ideology, which is different to tort. As the House of Lords once observed:
the second inquiry [after cause-in-fact], although this is not always openly acknowledged by the courts, involves a value judgement… The inquiry is whether the plaintiff's harm or loss should be within the scope of the defendant's liability, given the reasons why the law has recognised the cause of action in question.Footnote 162
Thus understanding the wrong allows us to explore the philosophical foundations underlying the cause of action. As the House of Lords noted, only after that can we draw conclusions about remedies.
Conclusion
Just like tort law, human rights law is an aggregate entity, composed of separate – but related – units. Those units deserve consideration individually. This paper makes a start on that project. The aim was to come away with a better understanding of the Osman cause of action by looking into the mirror of tort law; thereby understanding tort law better too. This paper shows why the Supreme Court's criticism of the ECtHR for ‘creating what amounts… to an independent substantive law of tort, overlapping with domestic tort law, but limited to cases involving death or the risk of death’Footnote 163 is unwarranted (and doctrinally incorrect because the PDP is not so limited). Theoretical clarity may not dispel antagonism by itself, but it ensures the antagonism is not misdirected. It does some heavy lifting to assist jobbing judges, and makes more transparent the political choices against which the ire is directed. To the extent that the debate about taxonomy is a proxy for deeper concerns, it forces us to be upfront about those concerns. The PDP does not fit in tort law. Further scholarship is needed on the other causes of action in human rights before we can conclude how to categorise ‘human rights law’ as a monolithic entity.
Examining the cause of action shows why the PDP takes a fundamentally different line to the law of negligence, despite superficial similarities. Much of it comes down to underlying ideas about the state. Turpin and Tomkins argue English law had no theory of the state until very recently,Footnote 164 but historically understood itself through the forms of action. This may explain why tort law treats individuals and public authorities alike. But the introduction of the Human Rights Act, 20 years ago, brought with it the discourse of a different legal tradition. It is important for tort lawyers and human rights lawyers, English lawyers and European lawyers, to deepen our understanding of the causes of action introduced by the HRA. For English lawyers, who have used the framework of the forms of action to understand our own law, this seems a good way in to the ECtHR's jurisprudence.
Certainly, it is more promising than an analysis of remedies or limitation periods, which can complement it, but tell us little in isolation. You might reasonably want to know about features other than remedy. Remedies may not tell us very much, if anything, about rights. The remedy might be serving a different function to the right. Indeed, remedies can serve dual functions. The problem is compounded by money having fluid symbolism. Crucially, remedies respond to something and it is the ‘something’ that I would want to know about.
It matters less that I am correct in my conceptualisation of the PDP and more that the conversation goes in the direction of examining HRA causes of action. Until shown otherwise, it seems to me that breaches of the PDP are wrongs only public authorities can commit. However, this is not a subset of tort, distinguished by the identity of the actor. How the wrong is committed and what makes it a wrong distinguishes it. The state did not stop something from happening. This failure, of itself, constituted a breach of a protected right. It is something the state should have done – an act of governance. It is this that was a violation of rights; it is not just that it led to a prohibited outcome. The state is not vicariously liable for the third party, or indemnifying citizens against crime. This becomes clear when we compare the structure of a complaint directed toward the third-party with a complaint directed at the state. The accusation is not ‘you hit me’, but rather ‘you did not stop him from hitting me’.
Interposed in the allegation directed toward the state is a sense of responsibility towards the claimant and for the perpetrator. Only the state is in a position to have both types of responsibility, towards and for diverse individuals. This gives the complainant the ingredients to say ‘you should have stopped him because you had a duty’, and goes to the philosophical foundations of the cause of action. English tort law is uncomfortable with both elements. Its discomfort with responsibility towards manifests in the ‘conferral of benefit’ doctrine. What tort regards as a matter of largesse, human rights regards as entitlement. Tort law's discomfort with responsibility for manifests in the acts/omissions (or positive/negative duty) distinction. This shades over into causation insofar as it relates to responsibility.
The natural retort of the state faced with a PDP allegation, ‘How should I have stopped him?’, can also be expanded into something slightly more sophisticated. It can be rephrased as ‘in what circumstances should I have stopped him and what should I have done?’ The answer then builds in the familiar parts of the Osman test, ie ‘where the authorities are aware or where they ought to have been aware…’ and so on. I have not tried, in this paper, to justify the particular formulation of the Osman testFootnote 165 but rather to understand what grounds the duty. It is in all our interests that we understand what the cause of action does. If we are convinced that it serves a valuable function by addressing a wrong which is not met by the tort of negligence then there is reason to allow it to stand as a parallel cause of action.
