Commissioner of Police of the Metropolis v DSD and another

Between 2003 and 2008, John Worboys, the driver of a black cab in London, committed a legion of sexual offences on women. The first respondent in these proceedings (who has been referred to throughout as DSD) was among his first victims. She was attacked in 2003. The second respondent (NBV) became Worboys’ victim in July 2007. Many others were attacked by him between 2003 and 2007 and, sadly, yet more after NBV was assaulted.

DSD and NBV brought proceedings against the Commissioner of the Metropolitan Police Service (MPS) for the alleged failure of the police to conduct effective investigations into Worboys’ crimes. The claims were brought under sections 7 and 8 of the Human Rights Act 1998 (HRA). The combined effect of these provisions (so far as this case is concerned) is to allow a person who claims that a public authority has acted in a way which is incompatible with their rights under the European Convention for the Protection of Human Rights and Fundamental Freedoms (ECHR) to bring proceedings against the public authority and to be awarded damages.

The kernel of DSD and NBV’s claims is that the police failures in the investigation of the crimes committed by Worboys constituted a violation of their rights under article 3 of ECHR. This provides that “[n]o one shall be subjected to torture or to inhuman or degrading treatment or punishment.” They succeeded in that claim before Green J, who delivered judgment on the liability issues on 28 February 2014, (Neutral Citation [2014] EWHC 436 (QB)). In a second judgment handed down on 23 July 2014, (Neutral Citation [2014] EWHC 2493 (QB)), Green J awarded compensation to DSD and NBV against MPS. An appeal by MPS was dismissed by the Court of Appeal (Lord Dyson MR, Laws and Kitchin LJJ) on 30 June 2015, (Neutral Citation [2015] EWCA Civ 646): [2016] QB 161). MPS has appealed to this court. The Secretary of State for the Home Department (SSHD) intervened, making written and oral submissions. A number of other parties intervened. Liberty intervened, as did jointly the organisations Rape Crisis, England and Wales, the End Violence Against Women Coalition, Southall Black Sisters and the Nia Project. They made helpful written submissions and valuable oral submissions.

In this appeal MPS accepts that both respondents were subjected to serious sexual assault by Worboys. It further accepts that there were significant errors by the police in each of the investigations into the crimes committed against them. MPS has said that, whatever the outcome of this appeal, MPS will not seek to recoup any of the compensation and consequential costs which have been paid.

DSD and NBV have recovered compensation from Worboys and each of them has received an award from the Criminal Injuries Compensation Authority (CICA).

It is accepted that HRA imposes a general duty to investigate ill-treatment amounting to a violation of article 3 of ECHR. The main area of dispute is the nature of that duty. That issue has a number of themes. They can be summarised as follows:

Is it a public duty or one owed to individual victims of the breach of article 3?

Is it a systems or an operational duty?

Does the duty to investigate breaches of article 3 in relation to a particular individual arise only when it is alleged that state authorities are complicit in the breach?

If the duty comprehends an obligation to investigate breaches of article 3, even if there is no involvement of state agents, is there a right to claim compensation against the state?

Should the fact that a victim can obtain redress against an offender or make a claim under the CICA scheme affect consideration of the availability of a right to compensation under HRA?

In this context, is it relevant that UK courts have, so far, refused to recognise a common-law duty of care on the police in relation to the manner in which officers prevent and investigate crime?

Finally, it is suggested that it would require the clearest statement in consistent decisions of the European Court of Human Rights (ECtHR) Grand Chamber to the effect that a positive duty was owed by the state to individuals who suffered treatment contrary to article 3 at the hands of another individual before holding that the investigative duty of the state was animated. If ECtHR jurisprudence is found to be less than clear, the appropriate course was to allow the government to deploy its arguments in Strasbourg - R (Ullah) v Special Adjudicator [2004] 2 AC 323.

Many of these themes overlap and blend into one another and it is impossible to consider them other than compendiously, not least because that is how they are treated in many of the relevant authorities, both domestic and European. It is necessary at the outset, however, to recognise that examination of the nature of the duty is a multi-faceted exercise.

The appellant argues that the duty of the police to investigate, detect and prosecute crime is of a communal nature; one which is owed to the public at large, not to individual citizens. It is submitted that the “general rule” is that no private law duty is owed to victims of crime. This is because it has been deemed that it would not be “fair, just and reasonable” to impose such a duty. The consistent theme of judicial decisions on the liability of the police at common law has been, the appellant suggests, that there are overwhelming public policy reasons that no such liability should be recognised. The Court of Appeal in the present case had emphasised the importance of consistency between the common law and HRA. It would be anomalous, the appellant contends, for there to be different bases of duty owed by the police at common law and under HRA. Many of the public policy considerations which militated against the recognition of such a duty at common law apply with equal force to the duty to investigate that arises under article 3 of HRA.

It is accepted that there is a duty to investigate allegations of ill-treatment which amount to a violation of article 3 but it is suggested that this duty can be enforced through the disciplinary regime, under the independent oversight of the Independent Police Complaints Commission. The existence of this regime, the appellant says, reflects the public nature of that duty.

Insofar as article 3 of ECHR imposes a positive obligation to respond to ill-treatment by a member of the public who is not a state agent, that obligation, the appellant submits, is to put in place the legal structures required to ensure that a proper inquiry can be conducted. It does not extend to the operational content of an individual inquiry. The investigative obligation in relation to individual cases arises only where state agents are complicit in the alleged ill-treatment.

The SSHD submits that the origin of the investigative duty is Assenov v Bulgaria (1998) 28 EHRR 652 where ECtHR stated that an effective official investigation was required where there was an arguable claim of serious ill-treatment “by the police or other such agents of the state unlawfully and in breach of article 3” - para 102. It is therefore argued that the principled foundation for the implication of an investigative duty is to underpin the effectiveness of the express prohibition set out in article 3. That prohibition could only apply, the SSHD argues, to agents of the state, not to private individuals.

The respondents’ riposte to these arguments is that the state has a duty under article 3 to conduct an effective investigation into crimes which involve serious violence to an individual. This is a positive, protective obligation to take measures designed to ensure that individuals within its jurisdiction are not subject to the treatment which article 3 prohibits. The duty is not an abstract one owed to the public at large but can be invoked by an individual who demonstrates that the state’s failure to fulfil its obligation has led to her or his suffering treatment prohibited by the article.

The respondents submit that the appellant’s argument about the need for consistency between the common-law position and the availability of a claim under HRA is, properly analysed, one of justiciability. They point out that the appellant accepts that there is a domestic law duty to investigate effectively serious criminal offences. It is accepted that there were several deficiencies in the investigation of these offences. The decision by Parliament to enact HRA effectively disposes of the issue of justiciability. The incorporation of ECHR into domestic law made available to an individual a remedy for a breach by the state of the article 3 protective obligation. This was entirely in line with the jurisprudence of ECtHR.

In any event, the respondents argue, the public policy considerations which have been held to underlie the exemption from liability at common law do not translate to the position under HRA. As Lord Brown of Eaton-under-Heywood held in Van Colle v Chief Constable of Herts Police; Smith v Chief Constable of Sussex Police [2009] AC 225, para 138, Convention claims have quite different aims from civil actions.

On the appellant’s argument that the duty of the state under article 3 is confined to putting in place legal structures to prohibit such ill-treatment as is forbidden by the article, the respondents claim that this is unsustainable, again in light of the decision in Van Colle. That decision was premised on the existence of an operational duty to protect against a real and immediate risk of serious violence.

In MC v Bulgaria (2005) 40 EHRR 20 the applicant complained that she had been raped by two men when she was 14 years old. The men in question were interviewed but it was concluded that they had not used threats or violence and there was no evidence of resistance on the part of the complainant. The district prosecutor therefore issued a decree terminating the proceedings. The complainant’s application to ECtHR rested on the twin assertions that Bulgarian law did not provide effective protection against rape and sexual abuse as only cases where the victim had actively resisted were prosecuted, and that the authorities had not properly investigated her allegations. It is important to recognise that the court found that the failure properly to investigate her allegations constituted violation of her rights under articles 3 and 8 of ECHR. As I shall discuss below, the appellant has concentrated on the first of the assertions made by the applicant in MC. The second aspect of her complaint - that there was not a proper investigation of her allegations is a distinct and unconnected ground on which the court decided that article 3 had been breached.

In para 151 of MC ECtHR observed that, in a number of cases, article 3 of ECHR gives rise to a positive obligation to conduct an official investigation. The court expressly said that such positive obligations “cannot be considered in principle to be limited solely to cases of ill-treatment by state agents”. It concluded that the authorities had failed to explore the surrounding circumstances - para 177. On that account, there was a violation of the state’s positive obligation under article 3 - para 187. This was a duty owed to the applicant personally and she was awarded compensation - para 194.

A significant passage from the court’s judgment is to be found in para 153:

The binary nature of the positive obligation arising under these articles was noted by Green J in para 163 of his judgment: effective systems and operational duties:

Lord Hughes has suggested (in para 117 of his judgment) that the statement in para 153 of ECtHR’s judgment, that “article 3 carries an obligation in some circumstances to investigate third party offending” leaves “only uncertainties about its source and thus its extent”. What is not in the least uncertain, however, is that, if the relevant circumstances are present, there is a duty on the part of state authorities to investigate where non-state agents are responsible for the infliction of the harm. That cannot be characterised as other than an operational duty. The debate must focus, therefore, not on the existence of such a duty but on the circumstances in which it is animated.

It is suggested (para 119 of Lord Hughes’ judgment) that Calvelli and Ciglio v Italy (2002) (Application No 32967/96) does not provide authority for the second proposition in para 151 of MC, namely, that positive obligations to investigate cannot be considered in principle to be limited solely to cases of ill-treatment by state agents. It should be observed, however, that part of the applicants’ complaint in that case related to the unexplained delay in the proper investigation of their representations that the doctor who was ultimately charged with manslaughter was responsible for the death of their child - see para 43 of the judgment. In other words, an operational failure. At para 54, the court said this:

The rejection of the applicants’ case in Calvelli and Ciglio was not because ECtHR considered that the duty to investigate was confined to “an obligation to provide a sufficient investigative structure”, as Lord Hughes puts it in the final sentence of para 119 of his judgment. To the contrary, the Strasbourg court held that, because of the applicants’ entitlement to issue proceedings in the civil courts and the fact that they entered into a settlement agreement with the doctor’s insurers, they had waived their rights to pursue criminal proceedings (para 54). Nothing in the court’s judgment supports the suggestion that it was founded on a view that the extent of the ancillary duty under article 2 was to provide a sufficient investigative structure rather than a duty not to be negligent in the way in which the inquiry was conducted. The court made that unequivocally clear in para 56 of its judgment where it said that because the applicants had denied themselves the “best means” of elucidating the extent of the doctor’s responsibility for the death of their child, it was unnecessary to examine whether the time bar on bringing proceedings prevented the doctor from being prosecuted. The coming into force of the time bar had, of course, been contributed to by the operational failure of the state authorities to conduct investigations more expeditiously.

Lord Hughes has described the statement in para 152 of MC as “tentative” - see para 121 of his judgment. It seems to me clear, however, that the court was there recording that, hitherto, ECtHR had not excluded the prospect that it would be held that actions by non-state agents would give rise to positive obligations on the part of the state under articles 2 and 3 of ECHR. MC provided the occasion to proclaim that such a positive obligation existed.

In order to be an effective deterrent, laws which prohibit conduct constituting a breach of article 3 must be rigorously enforced and complaints of such conduct must be properly investigated. There is a clear line of Strasbourg authority for the duty to properly investigate reported offences and allegations of ill-treatment, which is summarised with approval at para 172 of O’Keeffe v Ireland (35810/09):

It has been suggested (Lord Hughes at para 123) that ECtHR in MC made it clear that it regarded the deficiencies in the investigation as “the consequence of, and part and parcel with, the flawed approach of the Bulgarian system generally to the issue of lack of consent”. At para 179 the court said this:

In my view, the court was not suggesting in this passage that the deficiencies in the investigation were somehow subsumed into the shortcomings of the Bulgarian law. Nor was it suggested that deficiencies in investigation had to be accompanied in all circumstances by systemic defects. In MC, the lack of assiduity in the investigation could be explained because of the inadequacy of that law but it does not follow that deficiencies in investigation, if they are sufficiently egregious, cannot of themselves constitute a violation of article 3. Cases decided after MC make that unambiguously clear. I will discuss those cases presently.

Concentrating for the moment on MC, however, Lord Hughes suggests that his thesis, that deficiencies in investigation were part and parcel of “the flawed approach of the Bulgarian system generally”, was supported by the words of para 168 which he quotes at para 122 of his judgment. In para 168, ECtHR said that it was not concerned with “allegations of errors or isolated omissions in the investigation”. The court accepted that it could not replace the domestic authorities’ assessment of the facts of the case nor could it decide on the alleged perpetrators’ criminal responsibility.

These statements must be seen in context. The Strasbourg court is a supra-national body. There are obvious limitations on its opportunity to examine deficiencies in investigation. National courts are not so constrained. This case provides the perfect example. Green J heard detailed evidence of the errors that had been made by police in the investigation of Worboys’ crimes. He was in a position to form a judgment as to the impact of those errors on the respondents’ cases. And it was open to him to find, as he correctly did, that the errors were so serious that a violation of article 3 was established.

I cannot accept a suggestion that, to give rise to a breach of article 3, deficiencies in investigation had to be part and parcel of a flawed approach of the system generally. I accept, however, that simple errors or isolated omissions will not give rise to a violation of article 3 at the supra-national and the national levels. That is why, as I point out below, only conspicuous or substantial errors in investigation would qualify. The Strasbourg court disavowed any close examination of the errors in investigation because it was a supra-national court. It left that to national courts. But, my reference to ECtHR’s disinclination to conduct such a close examination is not intended to suggest that minor errors in investigation will give rise to a breach of the Convention right on the national plane. To the contrary, as I make clear in paras 53 and 72 below, errors in investigation, to give rise to a breach of article 3, must be egregious and significant.

As I hope is now clear, not every error in investigation will give rise to a breach of article 3. But the difficulty in defining those errors which qualify should not prompt capitulation to the notion that there has to be some form of structural deficiency before egregious errors in the investigation of the offences, such as occurred in this case, can amount to a breach of article 3. That proposition is strongly supported by consideration of cases decided after MC and I turn now to those cases.

The case of Szula v United Kingdom (2007) 44 EHRR SE19 involved a complaint of sexual and physical abuse brought by a minor during the time that he was in a residential approved school. The applicant’s claim was deemed inadmissible but this was because it was concluded that there was “no indication that the authorities showed any lack of diligence or expedition” in the investigation of his allegations. Implicit in that finding was that, had there been such an indication, the applicant’s case would have been admissible. It has been suggested that this case is an example of the court having looked for evidence of a structural defect or culpable disregard or an absence of good faith in the administration of the domestic system (para 26). I do not agree. In that case the court expressly recognised that the criminal law prohibited the physical and sexual abuse alleged by the applicant. True it is that the court, after reviewing steps taken by police and prosecuting authorities, also said (in para 1 of its judgment):

It is unquestionably clear that these observations were made in relation to the discharge by the police and the prosecuting authorities of their operational duties. There is no hint in the judgment that this was in any way related to a “structural defect”. As I have said, the court had examined the criminal law system and not found it wanting. I cannot accept, therefore, that the quoted passage had anything whatever to do with a systemic or structural failure. It was plainly pertinent - and only pertinent - to a review of the operational actions and decisions of the police and prosecuting authorities. The fact that the court considered it necessary to conduct such a review, when no systemic defect was present, is important, however. It can only have been necessary if the court considered that a purely operational failing, entirely unrelated to any “structural defect”, could have given rise to a violation of article 3.

In Secic v Croatia (2009) 49 EHRR 18 (31 May 2007), ECtHR considered a complaint of ineffective criminal investigation of a racially motivated physical assault. The court again repeated the statement from MC that article 3 may give rise to a positive obligation to conduct an official investigation - para 53. The court stated that the obligation on the state to conduct an official investigation is one of means, not result, referring to the article 2 cases of Menson v United Kingdom (2003) 37 EHRR CD 220 and Yasa v Turkey (1999) 28 EHRR 408. At para 54, however, it observed that the authorities had to take “all reasonable steps available to them to secure the evidence concerning the incident” and that the authorities must act with “promptness and reasonable expedition.” Having considered the investigations conducted by the police, ECtHR concluded, at para 59, that “the failure of the state authorities to further the case or obtain any tangible evidence with a view to identifying and arresting the attackers over a prolonged period of time indicates that the investigation did not meet the requirements of article 3 of the Convention”. It therefore found that article 3 had been breached and that the applicant was entitled to be compensated. Lord Hughes has cited this case as an example where there were “plain overtones of structural State deficiencies in relation to the investigation of allegations of racially motivated or discriminatory violence” (para 126).

The applicant’s submissions are set out in paras 38-42 of the court’s judgment. None of these submissions touched on any structural or systemic deficiency in Croatian law or procedure. To the contrary, the applicant complained that Croatian law provided for many processes and police methods which the police had failed to follow. He also complained that they had failed to apply to a Croatian court for an order compelling a journalist to reveal the name of an interviewee who might have been able to shed light on the attack on him. Such an application could have been made under section 30 of the Media Act. No complaint was made about any inadequacy in that provision. All of the applicant’s complaints were in relation to the operational failings of the police.

The Croatian government’s submissions are set out in paras 43-48 of the Strasbourg court’s judgment. Apart from claiming that the applicant’s ill-treatment did not reach the threshold required for a breach of article 3 and that the positive obligation under that article arose only where the state had been made aware of acts which it was reasonable to expect them to prevent, all the government’s submissions were focused on a defence of the operational decisions and actions of the police. The question of systemic deficiencies was simply not in play in this case. That is obvious also from the court’s decision. At para 53 the court said that article 3 may give rise to a positive obligation to conduct an official investigation. This was not, in principle, limited to cases of ill-treatment by state agents. And at para 54 the court said this:

The complaint of lack of promptness related solely to police inaction. Nothing about any structural or systemic deficiency was instanced. And the remainder of the court’s judgment focused entirely on the operational failings of the police. For my part, therefore, I have not been able to find any overtones of structural state deficiencies in the report of this case.

Beganović v Croatia (Application No 46423/06) 25 June 2009 was a case in which the applicant had been assaulted by three individuals. Although ECtHR acknowledged (in para 69 of its judgment) that no direct responsibility can attach to a member state under ECHR for the acts of private individuals, it stated (in paras 70 and 71):

The court made clear that, as well as examining the “impugned regulations and practices, and in particular the domestic authorities’ compliance with the relevant procedural rules”, it would also consider “the manner in which the criminal mechanisms were implemented in the instant case” - para 74. At para 75, ECtHR set out the “minimum standards applicable” in respect of “the duty to investigate”. They included that the investigation be independent, impartial and subject to public scrutiny, and that the authorities act with diligence and promptness. It also reiterated that “for an investigation to be considered effective, the authorities must take whatever reasonable steps they can to secure the evidence concerning the incident, including, inter alia, a detailed statement concerning the allegations from the alleged victim, eyewitness testimony, forensic evidence and, where appropriate, additional medical reports”. The failings in fact identified in this case arose at the post-investigative stage and ECtHR confirmed the principle that the requirement for effective criminal law provisions extends to the trial phase of proceedings - para 77.

The ECtHR decided that the state authorities did not fulfil their positive obligations under article 3. Violation of that article was found. Compensation was awarded to the applicant. The various elements of an effective investigation identified by the court should be noted. It must be independent. It requires to be prompt. Evidence must be secured. Failure to adhere to these standards renders the state liable to the individual affected by that failure.

In the case of Vasilyev v Russia (Application No 32704/04) 17 December 2009 the applicant and his friend were seriously assaulted and robbed. Although police officers attended the scene, no investigation into the circumstances of the assault were conducted. The police officers claimed to have considered that the applicant and his friend were intoxicated, so they moved them from the position where they had been found and left them. A number of criminal investigations were subsequently instituted, largely on the initiative of the applicant’s mother. It was decided to suspend the proceedings because the perpetrators could not be identified. This decision was reversed and restored on a number of occasions. The two police officers who had attended the scene were prosecuted for failing to fulfil their legal duty to protect victims of offences. They were acquitted.

The applicant did not lay blame on the state authorities for the attack; nor was it suggested that they knew or ought to have known that the applicant was at risk of physical violence at the hands of third parties. The court explicitly found, however, that this did not absolve the state from obligations under article 3. At para 99, it said that what the article required was that “the authorities conduct an effective investigation into the alleged ill-treatment even if such treatment has been inflicted by private individuals”. It elaborated on this statement at para 100:

It was held that there had been a violation of article 3 under its procedural limb in that the investigation into the assault on the applicant was ineffective. He was awarded compensation.

A similar approach to that in the cases already discussed is found in later decisions of ECtHR such as Milanovic v Serbia (Application No 44614/07) 14 December 2010, CAS v Romania (2015) 61 EHRR 18 and BV v Croatia (Application No 38435/13) 21 January 2016. The statement of the applicable principles concerning the procedural obligations in CAS v Romania (which reflected the exposition of those in the cases considered in detail above) was expressly endorsed by the Grand Chamber in O’Keefe v Ireland (2014) 59 EHRR 15. These propositions have been reiterated by the second section of ECtHR most recently in BV v Belgium (Application No 61030/08) 2 May 2017. At para 56 the court stated that the obligation to carry out an effective investigation “cannot be limited to cases of ill-treatment by agents of the state”.

In R (Ullah) v Special Adjudicator [2004] 2 AC 323, Lord Bingham of Cornhill, at para 20, quoted with approval the statement of Lord Slynn of Hadley in R (Alconbury Developments Ltd) v Secretary of State for the Environment, Transport and the Regions [2003] 2 AC 295, para 26, where he said:

The respondents argue that the authorities which I have reviewed above constitute clear and constant jurisprudence to the effect that the state has a duty under article 3 to conduct an effective investigation into crimes which involve serious violence to an individual. In order that the protective right under the article be practical and effective, the respondents also assert that failure to conduct such an investigation gives rise to a right to hold the state to account by, among other things, a claim for compensation on the part of a person who was a victim of the article 3 violation and the state’s failure to discharge its obligations under the article.

The appellant counters this argument, claiming that the Strasbourg court’s extensive case law (including decisions of the Grand Chamber) refers back to Assenov as the authoritative source of the obligation, derived from article 3, to investigate allegations of ill-treatment by state agents. It is suggested that Assenov explicitly limits the investigative duty to cases where the ill-treatment has been perpetrated “by the police or other such agents of the state unlawfully and in breach of article 3” (para 102). This approach, it is claimed, has been followed by decisions of the Grand Chamber in such cases as Gäfgen v Germany (2011) 52 EHRR 1 and El-Masri v The Former Yugoslav Republic of Macedonia (2012) 57 EHRR 25.

It is to be noted, of course, that Assenov was a case where the claim was based on allegations against state agents, namely, the Bulgarian police. When, therefore, the court expressed the view, which it did in para 102 of its judgment, that, “where an individual raises an arguable claim that he has been seriously ill-treated by the police or other such agents of the State unlawfully and in breach of article 3, that provision … requires by implication that there should be an effective official investigation”, it did not address the question whether such a duty arose where the perpetrators of the ill-treatment were not state agents. It did not need to do so. The issue simply did not arise in that case.

Likewise, in Gäfgen the complaint was made concerning the applicant’s ill-treatment by police. So also in El-Masri the applicant claimed that he had suffered ill-treatment at the hands of state agents and that they had been actively involved in his subsequent rendition by CIA agents. Statements in the judgments in both cases which echo that in Assenov quoted in the preceding paragraph do not sound on the question whether ill-treatment by individuals other than state agents can give rise to the duty to investigate under article 3. Those decisions do not impinge upon, much less derogate from, the authority of the cases decided by ECtHR between 2005 and 2017 discussed in paras 16 to 28 above. I consider that those cases amount to clear and constant case law of the Strasbourg court. And I have concluded that they establish that the state is obliged under article 3 to conduct an effective investigation into crimes which involve serious violence to persons, whether that has been carried out by state agents or individual criminals. Further, in order that the protective right should be practical and effective, an individual who has suffered ill-treatment contrary to article 3 has a right to claim compensation against the state where there has been a failure by state authorities to conduct a sufficient investigation into the crime.

At para 127 of Lord Hughes’ judgment, he suggests that the proper test for the positive obligation to investigate reports of past violence under article 3 is whether the state “has a proper structure of legal provision designed to punish it when it occurs and has administered [it] in good faith and with proper regard for the gravity of the behaviour under consideration”. It is not clear whether this formulation excludes any investigation of operational failings on the part of state authorities such as the police. Conceivably, I suppose, a failure to administer the proper structure of criminal legal provisions might entail an examination of the way in which police conducted their inquiries into a particular case. But, importantly, the duty to administer is qualified in Lord Hughes’ exposition by the stipulation that the administration of the structure of legal provision, if it is to fall foul of the test, must be shown to have been conducted in bad faith or without proper regard for the gravity of the behaviour involved. This places an obvious limitation on the scope of any review of the operational actions and decisions of the police. There is no suggestion in this case, for instance, that the police acted in bad faith. It might possibly be said that there were instances of the police failing to have regard for the gravity of the crimes which the respondents complained of but I am unsure whether this would fulfil Lord Hughes’ test because the failure to have due regard to the gravity of the crime must take place in the context of the administration of the proper structure of legal provisions.

It is true, as Lord Hughes says in para 140, that there were structural errors. But I cannot agree with his statement that the various detailed failings in the conduct of the inquiry were “largely attributable to this flawed structural approach”. Green J dealt with the operational failures in these two cases in a long passage of his judgment between paras 285 and 313. Some of these were related by him to lack of training but many were not. Significantly, the judge found that if the operational failings had not occurred, the police officers involved in the investigation “would have taken steps which would have been capable of identifying and arresting Worboys”.

It is unnecessary to list all the operational failings. These are set out in admirable and clear detail by the trial judge in his judgment. It is sufficient to refer to a sample of these to explain why I do not accept that these were largely attributable to a flawed structural approach.

None of these failures can be described as a failure in training or in the structures that were in place for the investigation of serious crime at the material time. Many other operational failures, none of which can be ascribed to a “flawed structural approach” were found by Green J to have occurred. These were considered by him to have contributed in a significant way to his finding that a breach of article 3 had been established. If I have understood Lord Hughes’ formulation of the relevant test correctly, none of them was relevant to that conclusion.

The prospect of every complaint of burglary, car theft or fraud becoming the subject of an action under the Human Rights Act has been raised. I do not believe that this is a serious possibility. All of the cases in this area involve conspicuous and substantial shortcomings in the conduct of the police and prosecutorial investigation. And, as this case illustrates, frequently, operational failures will be accompanied by systemic defects. The recognition that really serious operational failures by police in the investigation of offences can give rise to a breach of article 3 cannot realistically be said to herald an avalanche of claims for every retrospectively detected error in police investigations of minor crime.

The appellant argued that MC v Bulgaria should not be taken as authority for the proposition that how the state carried out its investigative duty at an operational level required to be examined in order to determine whether article 3 had been breached. That case, it was said, involved a systemic problem with Bulgarian law in relation to sexual offences. Under that law, it was not sufficient, in a rape prosecution, to show that a complainant had not consented to sexual intercourse. It was necessary to show that she was incapable of defending herself, or that she had been compelled by force or threats or that she had been brought to a state of defencelessness. The case was therefore primarily concerned with the system of laws in Bulgaria, the appellant claimed, and the court did not find a breach of article 3 because of any particular failing in the investigation in isolation, but because the legal system itself was deficient. The appellant claims that nothing in the judgment says in positive terms that article 3 gives rise to an obligation to investigate in cases where the state is not complicit in ill-treatment.

I do not accept these arguments. As pointed out in para 18 above, the Strasbourg court in MC clearly specified that the state’s duty had two aspects. The first was to enact criminal-law provisions which would effectively punish rape. The second, distinct but definite obligation was to carry out proper investigation and prosecution so that the laws could be applied effectively. It should be noted that the applicant’s complaint in that case had two separate aspects, described in para 109 of the judgment as follows:

The second aspect of her complaint was elaborated on in para 117 of the judgment where it is recorded that she alleged that the investigation had not been thorough and complete. The crucial issue of the timing of all the movements of the men and the applicant during the night in question had not been investigated.

The court’s conclusions on the second aspect of the applicant’s complaint were unmistakable. At paras 176-178 it said:

Plainly, therefore, the court made a separate finding in relation to the inadequacy of the police investigation. This finding was entirely freestanding of its conclusions in relation to the systemic deficiencies in the Bulgarian law in relation to rape. That approach has been consistently followed in the cases examined above. It is incontestably clear, therefore, that the positive obligation to conduct a proper inquiry into behaviour amounting to breach of article 3 may constitute a violation of the state’s duty under the article.

The answer to the argument that the positive obligation to investigate is animated only where there is state involvement in the acts said to breach article 3 can be simply supplied by reference to the passage from para 151 of MC quoted at para 17 above. The statement that positive obligations are not solely confined to cases of ill-treatment by state agents could not be clearer.

In fact, of course, statements to like effect appear repeatedly in ECtHR jurisprudence - see, for instance, para 70 of Beganović quoted at para 37 above; Vasilyev where the applicant expressly disavowed any accusation of blame on the state authorities for the attack on him; and para 83 of Milanovic (mentioned at para 43 above) where the court said:

Likewise, in CAS v Romania where there was no question of state involvement in the sexual abuse of the first applicant, the court was unambiguous in its finding that this was not a prerequisite to a breach of article 3. At para 69, it said:

I am satisfied, therefore, that ECtHR has consistently held that it is not required that there be state involvement in the acts alleged to amount to breach of article 3. The appellant’s argument based on that proposition must be rejected.

The themes outlined in para 6 (iv) and (v) above may be taken together. They can be dealt with briefly. Compensation is by no means automatically payable for breaches of the article 3 duty to investigate and prosecute crime. As Lord Bingham pointed out in R (Greenfield) v Secretary of State for the Home Department [2005] 1 WLR 673, para 8, in many cases the Strasbourg court has treated the finding of the violation as, in itself, just satisfaction under article 41 (although that was said in the context of article 6 breaches).

It is well settled, however, that the award of compensation for breach of a Convention right serves a purpose which is distinctly different from that of an order for the payment of damages in a civil action. As Lord Brown said in Van Colle at para 138:

Laws LJ said in para 68 of his judgment in the Court of Appeal, that the inquiry into compliance with the article 3 duty is “first and foremost concerned, not with the effect on the claimant, but with the overall nature of the investigative steps to be taken by the State”. I agree with that. The award of compensation is geared principally to the upholding of standards concerning the discharge of the state’s duty to conduct proper investigations into criminal conduct which falls foul of article 3. In paras 72-78 of his judgment, Laws LJ set out the systemic and operational failures of the appellant, quoting extensively from the judgment of Green J as to the first of these. That catalogue of failures was considered to warrant the award of compensation to the respondents, irrespective of the fact that they had received damages from both Worboys and CICA. I cannot find any flaw in the judge’s decision to award that compensation nor in the Court of Appeal’s decision to uphold that decision.

In Van Colle and Smith, two associated cases heard together, the complaint was that police had failed to follow up reports of threats to kill. In Van Colle, the alleged failure had resulted, it was claimed, in the killing of the individual who was the subject of the threats. In Smith, the victim was seriously injured. The first case was brought solely under HRA, alleging violation of article 2. It failed on its facts. In Smith, no HRA claim was made. The appellant relied solely on the common law, alleging negligence by the police. The House of Lords rejected the argument that “the common law should now be developed to reflect the Strasbourg jurisprudence about the positive obligation arising under articles 2 and 3 of the Convention” (para 136). A similar approach was taken by the majority in this court in Michael v Chief Constable of South Wales Police [2015] AC 1732.

As Laws LJ, in the Court of Appeal in this case, pointed out, the essence of the argument on behalf of the appellants in those cases was that the common law rule (that police owe “no general duty of care. … to identify or apprehend an unknown criminal, nor. … a duty of care to individual members of the public who might suffer injury through the criminal’s activities …” - Hill v Chief Constable of West Yorkshire Police [1989] AC 53) should be moderated so as to accommodate the ECHR - para 30. As he observed, the converse is contended for in this appeal. The appellant and the Secretary of State argue that the exemption from liability of the police at common law should be extended to claims advanced under HRA so that the two systems should be in harmony. There are two reasons for rejecting the argument.

In the first place, the bases of liability are different, as mentioned at para 44 above. In as much as it was considered that the common-law duty should not be adapted to harmonise with the perceived duty arising under ECHR, so should the latter duty remain free from the influence of the pre-HRA domestic law. Alternatively, it requires, at least, to be considered on its own merits, without the encumbrance of the corpus of jurisprudence under common-law.

Secondly and more importantly, no assumption should be made that the policy reasons which underlay the conclusion that an exemption of police from liability at common law apply mutatis mutandi to liability for breach of Convention rights. In Michael much of the debate as to whether police owed a duty to an individual member of the public centred on the question whether there was a sufficient proximity of relationship between the claimant and the police force against whom action was taken. No such considerations arise in the present context. The issue here is simple. Did the state through the police force fail to comply with its protective obligation under article 3?

The other principal argument advanced on behalf of the police in Michael was that it would not be “fair, just and reasonable” to impose liability on them for failings in individual cases. This is a concept with which the common law, with its innate flexibility, can cope but it is not one which can easily be accommodated in Convention jurisprudence. The police either have a protective duty under article 3 or they do not. The presence of the duty cannot depend on one’s conception of whether it is fair, just or reasonable for it to exist.

Lord Hughes has said (in para 130 of his judgment) that law enforcement and the investigation of crime involve a complex series of judgments and discretionary decisions; that they concern the choice of lines of inquiry, the weighing of evidence and the allocation of finite resources. All of that is unexceptionable. But the claim that to “re-visit such matters step-by-step by way of litigation … would inhibit the robust operation of police work … divert resources from current inquiries [and act as a deterrent] not a spur to law enforcement” is unsupported by any evidence. In the first place, none of the cases cited above required a painstaking, minute examination of decisions taken by police. Nothing in the Strasbourg jurisprudence suggests that this would be appropriate, much less that it would be even admissible, as the basis for advancing a claim under article 3. Carrying out police investigations efficiently should not give rise to a diversion of resources. On the contrary, it should lead to more effective investigation of crime, the enhancement of standards and the saving of resources. There is no reason to suppose that the existence of a right under article 3 to call to account egregious errors on the part of the police in the investigation of serious crime would do other than act as an incentive to avoid those errors and to deter, indeed eliminate, the making of such grievous mistakes.

The statement made by Lord Hughes (in para 130) about the undesirability of the investigation of terrorist activity and the “delicate and difficult decisions” it involves being subject to review would be a powerful factor, if it were a possible consequence of following the jurisprudence of ECtHR in this area. But, in my view, it is not. Nothing in that case law supports the notion that a charter has been created for the examination of every judgment or choice of strategy made. As I have said, only obvious and significant shortcomings in the conduct of the police and prosecutorial investigation will give rise to the possibility of a claim. There is no reason to suppose that courts will not be able to forestall challenges to police inquiries based on spurious or speculative claims.

It was strongly argued, particularly on behalf of the Secretary of State, that the question whether a liability such as that contended for by the respondents arises was one on which ECtHR should be invited to pronounce. The sub-text to this argument appeared to be that, where Strasbourg has not yet spoken, national courts should not venture forth.

This argument carries echoes of those which found favour in such cases as R (Al-Skeini) v Secretary of State for Defence (The Redress Trust intervening) [2007] UKHL 26; [2008] AC 153, and Ambrose v Harris Procurator Fiscal [2011] UKSC 43; [2011] 1 WLR 2435. In Al-Skeini Lord Brown suggested that where ECtHR had not spoken, our courts should hold back, explaining that, if it proved that Convention rights have been denied by too narrow a construction, the aggrieved individual can have the decision corrected in Strasbourg. And in R (Smith) v Oxfordshire Assistant Deputy Coroner (Equality and Human Rights Commission intervening) [2010] UKSC 29; [2011] 1 AC 1 Lord Phillips followed a similar line.

The difficulty with the argument is that it fails to address the circumstance that the courts of this country, constituted as they are as public authorities, must give effect to (or refuse to give effect to) Convention rights as a matter of domestic law. The HRA introduced to the law of the United Kingdom the European Convention on Human Rights and Fundamental Freedoms by making the Convention part of national law so that the rights became domestic rights. Because the rights are domestic, they must be given effect according to the correct interpretation of the domestic statute. As Lord Hoffmann said In re G (Adoption: Unmarried Couple) [2008] UKHL 38; [2009] AC 173, para 34 “[the courts’] first duty is to give effect to the domestic statute according to what they consider to be its proper meaning, even if its provisions are in the same language as the international instrument which is interpreted in Strasbourg”.

The so-called “mirror principle” (whereby pronouncements by national courts on Convention rights should precisely match those of Strasbourg) is often attributed to Lord Bingham’s statement in Ullah at para 20 where he said “The duty of national courts is to keep pace with the Strasbourg jurisprudence as it evolves over time: no more but certainly no less”. As explained in para 232 of R (Keyu) v Secretary of State for Foreign and Commonwealth Affairs [2015] UKSC 69; [2016] AC 1355, Lord Bingham was careful to refer to the interpretation of the Convention (as opposed to the interpretation of HRA). Despite this, his opinion in that case has been used in a number of subsequent judgments to support the proposition that the content of domestic rights under HRA should not, as a matter of principle, differ from that pronounced by Strasbourg. Indeed, his judgment has been construed as indicating that, unless ECtHR has given clear guidance on the nature and content of a particular Convention right, the national courts of the United Kingdom should refrain from recognising the substance of a claimed entitlement under ECHR - see, for instance, Al-Skeini, Smith and Ambrose, referred to in para 74 above.

In more recent cases, a departure from the mirror principle can be detected. Thus, in Rabone v Pennine Care NHS Foundation Trust (INQUEST intervening) [2012] UKSC 2; [2012] 2 AC 72 it was held that there was a positive obligation to protect the life of a mentally ill young woman who had been admitted to hospital informally because of serious attempts to take her own life. This decision was reached notwithstanding the fact that there was no authority from ECtHR to that effect. In Surrey County Council v P (Equality and Human Rights Commission intervening) [2014] UKSC 19; [2014] AC 896, para 62 Lord Neuberger said that where there was no Strasbourg authority which dealt precisely with the issues before this court, this court could rely on principles expressed by ECtHR, even if only indirectly relevant, and apply them to the cases which it had to decide. And in Moohan v Lord Advocate (Advocate General for Scotland intervening) [2014] UKSC 67; [2015] AC 901 Lord Wilson suggested that there had been a “retreat” from the Ullah principle which had led the court to “substantially” modify it. At para 105 he said:

This seems to me to be inescapably correct. Reticence by the courts of the UK to decide whether a Convention right has been violated would be an abnegation of our statutory obligation under section 6 of HRA. This section makes it unlawful for a public authority, including a court, to act in a way which is incompatible with a Convention right.

As it happens, of course, I consider that the jurisprudence of the Strasbourg court is clear and constant on the issues which this court has to decide. Even if it were not, however, I would firmly reject the suggestion that the decision of this court on whether the respondents enjoy a right under the HRA to claim compensation against the appellant should be influenced, much less inhibited, by any perceived absence of authoritative guidance from ECtHR.

For these reasons and for those given in the judgment of Lord Neuberger, with which I agree, I would dismiss the appeal.

The claimants, DSD and NBV, succeeded before Green J in establishing that they were entitled to damages from the defendant, the Commissioner of the Police of the Metropolis, as a result of failures by the police properly to investigate serious sexual assaults which had been perpetrated against them. The claims were founded on the propositions that (i) article 3 of the European Convention on Human Rights carries with it an obligation on the state to carry out an effective investigation when it receives a credible allegation that serious harm has been caused to an individual, and (ii) there were serious defects in the police investigation of the assaults on the claimants.

The Court of Appeal upheld the decision, and this court takes the same view. However, we disagree on one issue. That issue is whether a person in the claimants’ position needs to establish that the serious defects in the investigation in question were attributable to failures of a structural nature (also referred to as systems, or systemic, failures), and not to purely operational failures (ie failings on the part of the individual police officers responsible for conducting the specific investigation).

We do not need to decide this issue in order to resolve the appeal, but I agree that we should do so. It has been fully and helpfully argued by the parties and the interveners, the competing arguments have been admirably expounded by Lord Kerr and Lord Hughes, and it seems to me to be an important issue which should be decided if possible.

The competing arguments have been fully set out in the judgments of Lord Kerr and Lord Hughes. Lord Kerr favours the wider approach, namely that a claimant need only establish serious defects in the investigation into her particular case, irrespective of whether they are systemic or operational failures. Lord Hughes prefers the narrower approach, the effect of which is that a claimant has to establish serious failings of a systemic nature, and that failings of a purely operational nature will not suffice, at least where the perpetrator of the alleged assault was not a state agent.

In agreement with Lord Kerr, I am of the view, that serious failures which are purely operational will suffice to establish a claim that an investigation carried out pursuant to an article 3 (or indeed an article 2) duty infringed that duty.

So far as the Strasbourg jurisprudence is concerned, I consider that the judgments to which we have been referred support the wider approach. The investigatory duty was identified in Assenov v Bulgaria (1998) 28 EHRR 652, para 102, where the court said that article 3, read with article 1, “requires by implication that there should be an effective official investigation” into cases where “an individual raises an arguable claim that he has been seriously ill-treated by the police or other such agents of the state unlawfully and in breach of article 3”. I also note that the wider approach appears to have been adopted by the Grand Chamber in the article 2 case of Mustafa Tunç and Fecire Tunç v Turkey [2015] ECHR 383 - see at paras 183-209.

I accept of course that those decisions were concerned with cases of ill-treatment by state agents, and that the approval of the principle in other decisions of the Grand Chamber (eg Gäfgen v Germany (2010) 52 EHRR 1 and El Masri v The Former Yugoslav Republic of Macedonia (2012) 57 EHRR 25) were also concerned with such cases. I also accept that the standard which a court should apply when considering whether the investigatory duty has been satisfied may well be more stringent in cases where the alleged ill-treatment was caused by state agents than in cases where it was caused by others. However, I see no basis for holding that the duty is different in kind in the two types of case. While in a number of Strasbourg court decisions, it is stated that the duty extends to cases where the alleged ill-treatment was caused by third parties, there is no suggestion that the nature of the duty to investigate is different in kind in the two types of case. Thus, in the recent case of BV v Belgium (Application No 61030/08) 2 May 2017, para 56 the court stated that the obligation to carry out an effective investigation “cannot be limited to cases of ill-treatment by agents of the state”, without suggesting that there was any difference in the basic nature of the duty.

It is true that in Beganović v Croatia [2009] ECHR 992, para 69, the court said that “the scope of the State’s positive obligations might differ between cases where treatment contrary to article 3 of the Convention has been inflicted through the involvement of state agents and cases where violence is inflicted by private individuals”. However, even without considering other decisions of the Strasbourg court, that seems to me to be a very small peg on which to hang a contention that the two cases require approaches which differ in nature. In any event, when that observation was repeated in Vasilyev v Russia [2009] ECHR 2078, para 100, the court immediately went on to make the point that “the requirements as to an official investigation are similar”, and that was a point repeated in a number of subsequent decisions - see eg Koky v Slovakia [2012] ECHR 994, para 215, Amadayev v Russia [2014] ECHR 704, para 70, and MC and AC v Romania [2016] ECHR 359, para 111. Indeed, I note that in Beganović itself, when considering whether the investigatory duty had been complied with in a case where the alleged perpetrator had been a non-state agent, the court said in para 74 that it should consider “whether or not the impugned regulations and practices, and in particular the domestic authorities’ compliance with the relevant procedural rules, as well as the manner in which the criminal-law mechanisms were implemented in the instant case, were defective to the point of constituting a violation of the respondent state’s positive obligations under article 3”. And the subsequent analysis of the facts in paras 80-86, which led the court to conclude that there had been a violation in that case, focussed very much on the operational failures.

That approach appears to be supported by other Strasbourg court decisions involving the investigatory duty in relation to acts of serious ill-treatment by non-state agents. There is no suggestion in MC v Bulgaria (2005) 40 EHRR 20 that that duty is restricted to having effective systems in place: at para 153, the court referred to the duty of a state “to enact criminal-law provisions effectively punishing rape and to apply them in practice through effective investigation and prosecution”. In para 167, the court seems to me to have been referring to both systemic and operational failures when it mentioned “significant flaws” in “the impugned legislation and practice and its application in the case at hand, combined with the alleged shortcomings in the investigation” (and see para 179). The finding of inadmissibility in Szula v United Kingdom (2007) 44 EHRR SE19 appears to me to have been based on the assumption that operational failures would, in principle, suffice to found a claim - see para 1.

To the same effect, in Šečić v Croatia (2007) 49 EHRR 18, the court said in para 54 that the police should have taken “all reasonable steps available to them” to obtain relevant evidence, which, as a matter of ordinary language, naturally refers to operational steps. In para 59 the court concluded that what appear to me to have been operational, rather than structural, failures (summarised in paras 56-58) enabled the article 3 claim to succeed. In Beganović, para 75, the court made the point that “the duty to investigate” involved “the authorities [having to] take whatever reasonable steps they can to secure the evidence concerning the incident” - which covers operational matters at least as much as systems. Vasilyev is another case where the court described the duty in terms which are, at least in my view, significantly more consistent with the wider approach, namely that “the authorities [should] conduct an effective investigation” involving “the reasonable steps available to them” (paras 99 and 100). In the 2016 case of Chernaya v Ukraine [2016] ECHR 1117, which involved an injury inflicted by a non-state agent, the court reiterated in para 25 that “[t]he minimum standards of effectiveness laid down by the Court’s case law include the requirements that the investigation be independent, impartial and subject to public scrutiny, and that the competent authorities must act with exemplary diligence and promptness” - again focusing on the operational aspects of the particular investigation.

Of course, this Court is not required to follow Strasbourg jurisprudence, even in a case such as this where there is a clear and consistent approach adopted in a significant number of chamber decisions. Dialogue between the United Kingdom Supreme Court (and indeed other courts in the United Kingdom) and the Strasbourg court has proved to be beneficial to the development of human rights law in this jurisdiction - and, I hope, in Strasbourg. Accordingly, if it appears to us that the narrower approach is even only probably correct, the fact that the Strasbourg court has consistently taken a different view should not necessarily stand in the way of our coming to a contrary conclusion. In this case, the notion that we can take such a course can fairly be said to be supported by the fact that, although it is inconsistent with the views expressed in a number of decisions of the Strasbourg court, the notion that the narrower approach is correct has not, so far as I can see, been specifically raised in that court. But there must be a good reason for our taking such a course, and in this case, at least in my view, there is not.

Indeed, in my view, there are good reasons for favouring the wider approach. First, one starts with the proposition that, given that it is rightly accepted on all sides that the authorities have an investigatory duty, it would be of little value unless it was a duty to investigate effectively. Provided that courts bear clearly in mind “the difficulties involved in policing modern societies, the unpredictability of human conduct and the operational choices which must be made in terms of priorities and resources” and the need to interpret the duty “in a way which does not impose an impossible or disproportionate burden on the authorities” (Osman v United Kingdom (1998) 29 EHRR 245, para 116), I find it hard to understand why an investigation which is seriously defective in purely operational terms should, in effect, be held to satisfy the investigatory duty.

Secondly, I cannot see any basis in its jurisprudence to suggest that it is likely that the Strasbourg court would think it right to limit the extent of the investigatory duty to systemic, as opposed to operational, failures. It is true that in A v United Kingdom (1998) 27 EHRR 611, having held that article 3, together with article 1, “requires States to take measures designed to ensure that individuals within their jurisdiction are not subjected to torture or inhuman or degrading treatment or punishment”, the court concluded that then-current statutory provisions “did not provide adequate protection to the applicant against treatment or punishment contrary to article 3” and that “the failure to provide adequate protection constitutes a violation of article 3” (paras 21 and 24). However, that conclusion merely reflected the factual basis and arguments in the case. If the statute did not provide adequate protection, there was a systemic failure, and it was both unnecessary and pointless to consider the operational aspects of the legal system. However, clearly to my mind, it does not follow that, if the statutory provisions had complied with article 3, but the legal processes had been defective, the United Kingdom would have been acquitted of infringing article 3. Indeed, the court’s summary of the law in A v United Kingdom, para 22 contains nothing to suggest that the state’s obligation there being discussed should be limited to systemic matters.

Similarly, while the court in Osman, para 116, was concerned to ensure that the investigatory duty was not interpreted or applied unrealistically, there is no indication in that paragraph that it was intending to limit the duty to the provision of a satisfactory framework, irrespective of how ineptly it operated in a particular case. Indeed, such an approach would seem to me to be inconsistent with how the Strasbourg court approaches cases generally, namely by reference to the specific facts of the particular case.

Thirdly, there are forensic considerations. In that connection, I would start by rejecting the notion that it could be right for a court to dismiss a claim that an investigation was seriously defective simply because the relevant police procedures as set out in official documents were satisfactory. It would not merely be formalistic, but both unjust and unrealistic, to hold that an investigation, which was seriously systematically defective in practice, nonetheless complied with the article 3 investigatory duty simply on the grounds that, while the systemic defects occurred in practice, they did not reflect the systems as laid down officially. Whether the wider or the narrower approach is correct, the court must surely consider the real, not the hypothetical.

Once that is accepted, I consider that the narrower approach could present a court with difficult practical, categorisation, and apportionment issues. Whichever approach applies, a court must inevitably start by considering the failures in the particular case. On the wider approach, the court would simply ask whether those failures were sufficiently serious to represent an infringement of the investigatory duty. On the other hand, on the narrower approach, the court would have to consider which of the failures were operational and which were systemic, and that, as I see it, is where problems would often start. Serious operational failures by individual officers would frequently throw up arguable systemic issues, such as systems of supervision or even of appointment of those officers. And, in order to decide whether the operational failures were systemic in origin, the court might often have to embark on an inquiry whether, for instance, the failures were redolent of what happened in other investigations. That could involve a potentially time-consuming and expensive inquiry into other investigations, as well as arguments as to the number and types of investigation, if any, to which the inquiry should be restricted. The question whether the defective investigation was attributable to systemic, rather than purely operational, failures could also involve difficult issues of categorisation and inference. For instance, in many cases it may be hard to decide whether a particular failure is operational or systemic, or whether the operational failures in an investigation or a set of investigations entitle the court to infer a systemic failure. And what happens if, as may very often be the case, there are some operational failures which are purely operational and some which are attributable to structural failures?

I do not consider that my view is undermined by the reasoning expressed or conclusions reached in Hill v Chief Constable of West Yorkshire Police [1989] 1 AC 53, Brooks v Comr for the Police for the Metropolis [2005] 1 WLR 1495, Van Colle v Chief Constable of Hertfordshire Police; Smith v Chief Constable of Sussex Police [2009] AC 225 and Michael v Chief Constable of South Wales Police [2015] AC 1732. Those cases establish that, absent special factors, our domestic law adopts the view that, when investigating crime, the police owe no duty of care in tort to individual citizens. That is because courts in this country consider that the imposition of such a duty would, as Lord Hughes puts it, “inhibit the robust operation of police work, and divert resources from current inquiries; it would be detrimental, not a spur, to law enforcement”. That view is entirely defensible, but, at least in the absence of concrete evidence to the contrary, so is the opposite view that the imposition of such a duty, provided that it is realistically interpreted and applied, would serve to enhance the effectiveness of police operations. It is therefore understandable that human rights law, with its investigatory duty under article 2 and 3, differs from domestic tort law in holding that it is right to impose an investigatory duty on the police. Just as the majority of this Court accepted in Michael, at paras 123-128, that the domestic tortious test for liability should not be widened to achieve consistency with the human rights test, so should the human rights test for liability not be narrowed to achieve consistency with the domestic, tortious test.

Finally, I turn to an argument which I have already touched on, namely that the wider interpretation involves placing too great a burden on public authorities, in particular on the police. This concern was recognised in relation to the similar article 2 obligation on the police in Osman v United Kingdom (1998) 29 EHRR 245, para 116, where, as I have mentioned, the Strasbourg court said that the “obligation must be interpreted in a way which does not impose an impossible or disproportionate burden on the authorities” (and see to the same effect CN v United Kingdom (2012) 56 EHRR 24, para 68). The point was developed in MC v Bulgaria, where at para 168, the Strasbourg court made it clear that, when considering whether an investigation had satisfied article 2, a court should not be “concerned with allegations of errors or isolated omissions in the investigation”. As the Grand Chamber said in Tunç v Turkey, para 176, “the nature and degree of scrutiny which satisfy the minimum threshold of the investigation’s effectiveness depend on the circumstances of the particular case”. The court in that case went on to find “shortcomings” in the investigation, but held that there had been no infringement of article 2, because they were not “serious” or “decisive” (paras 189 and 195), and it concluded in para 209 that there were “no such shortcomings as might call into question the overall adequacy and promptness of the investigation”. And in Beganović v Croatia, para 78, the Strasbourg court emphasised that it should only conclude that the investigatory duty had been infringed “in cases of manifest disproportion between the gravity of the act and the results obtained at domestic level” (citing the earlier decisions of Nikolova and Velichkova v Bulgaria, (2009) 48 EHRR 40, para 62, and Atalay v Turkey [2008] ECHR 850, para 40, which are to the like effect). It is because of the concern expressed in Osman, para 116, that it is important to emphasise that only “serious” defects in any investigation can lead to the conclusion that there has been an infringement of article 2 or article 3.

Accordingly, I conclude that a claim based on serious defects in the performance of the investigatory duty under article 3 (and equally under article 2) can succeed even if defects are all of a purely operational, as opposed to a systemic, nature.

For these reasons, which are little more than a summary of those given by Lord Kerr, with whose judgment I agree, I would therefore dismiss this appeal.

I agree that the appeal of the Metropolitan Police in this case should be dismissed. It seems to me, however, important that the ambit of the positive duties which arise under article 3 ECHR should be subject to examination, if not that an exhaustive definition should be attempted.

The context in which this case comes to this court needs to be remarked. It is unusual. The treatment under consideration is the very serious offence of rape of victims who were exploited after putting themselves in apparently trustworthy hands (a black cab driver) and who had then been rendered helpless by stupefaction. There has been no dispute from the beginning that there were notable general failings in the police approach to investigation into the kind of complaint which both claimants made. They affected both the initial complaint by DSD in 2003, at least from the time of the morning after the event, at which stage she first realised that she had been sexually attacked, and the later complaint by NBV in 2007. The findings recorded by the judge in a meticulous judgment were almost entirely to the same effect as, and were grounded upon, the police’s own conclusions, following detailed internal reports, about the deficiencies of their approach to the possibility of drug induced rape. The consequences were extremely serious. Because the first incident reported was not approached properly, the attacker remained undetected and became a serial rapist. Although only a small proportion (ten) of his many attacks were reported to the police at the time, it is now known that he raped more than 100 women, employing a similar method. His modus operandi was highly specific. Once anyone put two or more of the reported incidents side by side, the inference that there was a single serial offender was irresistible. Once that was done, early in 2008, the rapist was arrested, and compelling evidence against him found, within eight days. That could and should have been done years earlier. Recognising the substantial justice of the complaints, the police have made it clear that in the event that their appeal should be successful, they do not seek repayment of the compensation ordered by the judge. But the elementary justice of the complainants’ cases makes it all the more important that the ambit of the duty should be considered. Otherwise indignation at their experiences may lead to an over-wide formulation passing unnoticed, with detrimental results for the criminal justice system.

Article 3 ECHR says this:

It is elementary that, as Laws LJ pointed out in the Court of Appeal in the present case, the obligations created by the Convention lie upon the party States and upon no one else. The Convention governs the relationship between the State and the citizen. It creates no duties for individuals. Article 1 is explicit. It is on the states that it imposes the obligation to “secure” to those within their jurisdiction the rights and freedoms defined in the Convention. It is perfectly clear that the primary case of behaviour which is a breach of article 3 is where torture or inhuman or degrading treatment is meted out by the state against a person within its jurisdiction. Even without the origins of the Convention in the aftermath of the Second World War, the very use of the expressions “torture” and “inhuman or degrading treatment” is sufficient demonstration of that; these are typically descriptions of State misbehaviour. And the same is demonstrated by the words “subjected to”. If the State inflicts such treatment, it has subjected the citizen to it. Anything beyond that is a judicial gloss on the Convention, well established as that gloss may now be. In fact, there have been developed two glosses. Similar glosses have been applied to the primary obligation in article 2, concerning the right to life, and it may be to other rights.

The first gloss was explained by the Strasbourg court in Assenov v Bulgaria (1998) 28 EHRR 652, in which there was an allegation of police violence towards a suspect in custody. It builds on the primary obligation of the state not itself to inflict prohibited treatment on the individual. If that primary obligation stood alone, it might well be ineffective because the State organ which inflicts such treatment may deny it or cover it up. It will be effective only if, when there is a reason to think that such ill-treatment may have been inflicted by a state organ, there is an ancillary “positive obligation” to take steps to investigate the allegation and to bring to book those who are found to be responsible. At para 102 the court set out the reason for this obligation:

The same rationale was explained (in the context of the ancillary article 2 positive obligation to investigate State responsibility for death in custody) by Lord Bingham in R (Amin) v Secretary of State for the Home Department [2004] 1 AC 653, para 31. The purpose of the duty is, he said:

The second gloss is different and is the one in question in the present case. It concerns ill-treatment not by the state but by an individual or other third party actor. Plainly a citizen may be subjected to inhuman or degrading treatment (or torture) by a fellow citizen. It has become commonplace to speak of this as a breach of article 3, but it is not. An individual cannot be in breach of the Convention. In the ordinary way, if treatment falling within article 3 is applied by A to B, then A may have subjected B to it, but the state has not. The State’s responsibility is simply not engaged. But it may be if, and only if, the State fails to take steps to afford its protection to B against A’s misbehaviour. In that event, it is still not a case of the state subjecting the individual to inhuman or degrading treatment, but by judicial gloss the article is read as carrying with it an unspoken but implicit “positive obligation” upon the State to afford its protection.

This second gloss has the potential to extend considerably wider than the first. Whilst the first is concerned to give effect to the primary obligation of the state not itself to subject people to inhuman or degrading treatment, the second reaches into the question of what the state is bound to do in relation to the acts of people for whose behaviour it is not responsible. When one comes to this question, there are considerations which do not apply to the first gloss.

Even in the most law-abiding of states, that sometimes serious harm will be inflicted by one individual upon another, in the context of all manner of disputes and in pursuit of many different objectives, is a regrettably unavoidable feature of life. No one suggests that the state is bound to guarantee that this will not happen. Indeed, some steps which an authoritarian state might be inclined to take with a view to preventing it (such as preventive detention without conviction or other court order, house arrest, intensive surveillance and the like) might themselves be infringements of other fundamental rights afforded to the citizen. That consideration apart, the systems which states adopt for the protection of those within their boundaries do not have to be the same. It has never been suggested that it is the function of the Convention to monitor every act of enforcement or policing of the varied domestic legal requirements, nor the content of those requirements, so long as they provide sufficiently for the protection of the individual against third party behaviour which meets the high threshold of severity contemplated by article 3. Otherwise, what would be involved would not be a gloss on the primary obligation imposed by article 3, nor would the “positive obligation” be in any sense ancillary to that primary obligation. Rather, the duty would be of a completely different character to the primary obligation, and would entail wholesale assimilation of, and judicial control of, the legal systems of independent states.

The positive obligation constituted by this second gloss - relating to the protection of citizens from third party ill-treatment - requires first that the state have a legal framework for the prohibition of conduct passing the article 3 threshold, and thus afford the protection of its legal system against such behaviour. A v United Kingdom (1998) 27 EHRR 611 was an example of a legal system which failed to provide sufficient protection to children against the infliction of serious harm via corporal punishment, until the former English defence of “reasonable chastisement” was altered. Such cases concern the structure of a state’s system.

The more difficult question is whether the implied positive obligation recognised by the second gloss extends beyond the structure of the state’s system to its operation in an individual case. The Strasbourg court confronted this in Osman v United Kingdom (1998) 29 EHRR 245, where the complaint was that the police force had failed sufficiently to heed reports suggesting that an obsessive might be a danger to an individual, who had then been attacked and almost killed (and his father killed). The context was the equivalent gloss on article 2, but the principles are the same as for article 3. The court concluded that the implied positive obligation to protect could apply but in narrow circumstances of “a real and immediate risk to the life of an identified individual by a third party, of which risk the State either knew or ought to have known”. If such an immediate threat exists, then the State’s obligation is to do what can reasonably be expected of it which might reasonably have avoided the risk; it is not limited to a duty to avoid gross negligence. As Lord Bingham observed in Van Colle v Chief Constable of Hertfordshire Police; Smith v Chief Constable of Sussex Police [2008] UKHL 50; [2009] AC 225 at para 29, it is quite apparent that every ingredient of this carefully drafted test is of importance. It defines the restricted circumstances in which a duty arises under the Convention to take reasonable operational steps to forestall known specific threats to the safety of an individual from eventuating.

The court in Osman made clear the reason why this test is restrictive. It lies in the realities of law enforcement and policing. At para 116 it said this:

This duty recognised in Osman is a duty to take reasonable steps not to enforce the law or to punish a perpetrator, but to prevent serious violence from occurring to an individual when the threat of it is sufficiently specific. It is an expansion of the second gloss beyond the requirement for structures and systems to render serious violence unlawful, but it is carefully narrow in extent. The House of Lords faithfully applied Osman in Van Colle v Chief Constable of Hertfordshire Police.

What has happened since is that the second gloss has been further extended beyond prevention of anticipated violent crime to a duty to investigate reported past violence. The usually identified origin of this further extension is MC v Bulgaria (2005) 40 EHRR 20. There is no doubt that the formulation adopted in that case has been often repeated, usually word for word, in later cases. In that sense there is, no doubt, a consistent line of authority. The difficulty is that much remains unclear, indeed unexplained. Unclear remain the rationale for the extension, the issue whether it can be derived from Osman and/or Assenov or if not why departure from the principles of those cases is justified, and the extent of the duty as extended.

MC v Bulgaria concerned a complaint of rape made by a 14 year old against two young men of her acquaintance with whom she had spent an evening and night travelling around in their car. They had been arrested on the day that she made her complaint to the police, a few days after the alleged event. Their case was that she had consented. There were multiple clashes of evidence between the complainant on the one hand and those she accused, plus other witnesses, on the other. But the prosecutors’ investigation had been closed on the grounds that neither force (physical or psychological) nor threats of the same, nor physical resistance by the complainant, had been established. The point of the case was the court’s rejection of that criterion for rape, which was at the time consistently adopted by the Bulgarian authorities, and was asserted by the Government to be the rule: see paras 122 and 166. Thus there was a breach of article 3 (and indeed of article 8) because the Bulgarian structures or system for the criminalisation of rape did not sufficiently protect an individual’s sexual autonomy; absence of consent must be the criterion, rather than the use of force.

That sufficiently demonstrated a breach of article 3 in the same way as in A v United Kingdom. The court, however, went on to draw attention to deficiencies in the prosecutors’ investigation of the complaint. At the outset of its judgment it made the following statement of principle, which has subsequently been adopted, often word for word, in later cases.

Whilst these paragraphs show that the court asserts that article 3 carries an obligation in some circumstances to investigate third party offending, they leave only uncertainties about its source and thus its extent.

As has been seen, Assenov does indeed show that article 3 gives rise to a positive obligation “in a number of cases” to conduct an official investigation. But those cases are ones where the investigation is into suspected state involvement in ill-treatment and is necessary if the primary obligation on the State is not to be ineffective. That rationale does not apply to cases of third party offending and MC v Bulgaria does not say anything about the difference between the two situations. Given the explanation in Assenov of the reasons why an ancillary positive obligation has been devised by judicial gloss, the difference is fundamental.

Calvelli and Ciglio was a case of alleged medical negligence in the course of the delivery of a baby who died shortly afterwards. It had been treated by the Italian authorities as a case of possible criminal manslaughter. The doctor had been convicted but his conviction was eventually set aside by the Court of Cassation on the grounds that he had wrongly been convicted in his absence. A re-trial was ordered but by then the limitation period had expired. The claimants asserted that there had been no effective system of investigation and trial because owing to delays the limitation period had expired and the doctor had not been convicted. The Court held that there had been no violation of article 2. The obligation discussed was the duty to run a system which provided sufficient remedies: see paras 49-54. That is the A v United Kingdom duty. There had been no breach of it because in the field of medical negligence a civil rather than a criminal process can suffice and here there had been sufficient remedies in (1) a right to damages (which had been pursued and then settled by the claimants without a finding of liability against the doctor, thus waiving an entitlement to pursue the case to such a finding) and (2) disciplinary regulation of doctors. True, there was complaint of delay made, but there was no examination in this case of the question what duty existed to investigate the doctor’s actions. The nearest that the court got to that question was the observation that remedies must be effective and not exist in theory only. It is not easy to see how this case generated the second sentence of para 151 in MC v Bulgaria, but if it did it is more consistent with an obligation to provide a sufficient investigative structure than with a duty not to be negligent in the detailed inquiry.

As to para 152 of MC v Bulgaria, para 128 of Osman (a) is concerned with whether there had been a breach of article 8 and (b) simply re-states the finding that the police did not know, nor ought to have known, of a real and immediate threat to the safety of the complainant. Of course it is true that in order to establish whether a sufficiently real and immediate threat to the safety of the complainant exists, it will in some cases be necessary for the police to investigate the complaint. But that does not alter the fact that the carefully limited duty recognised in Osman is concerned not with an obligation to investigate a past event, but with an obligation to prevent a future one.

The tentative tone of para 152 is also to be noted. No more is said than that the possibility of the positive obligation extending to a duty to mount an effective investigation is “not excluded”. No further reasoning is shown for the progression from that to the much more positive statement in para 153. It is not clear why that more positive statement follows from the previous two paragraphs, even if those two had been themselves firmly grounded on the authority cited.

The court in MC v Bulgaria was moreover at pains to set some limit to the permissible review of the investigation. At paras 167-168 it said this:

The court identified, at para 177, deficiencies in the investigation. It concluded that the conflicting assertions were not sufficiently sensitively assessed, and that inquiries which could have been made into timings, which might have shown which version was correct, were not made. It also criticised the fact that the complainant had not had the opportunity to confront and question the witnesses relied upon by the accused. Lastly, the investigators had not, it was said, sufficiently taken into account the unlikelihood that a 14 year old would make advances to the second man only minutes after losing her virginity to the first. But at para 179, set out by Lord Kerr at para 25 above, the court made it clear that it regarded the deficiencies in the investigation as the consequence of, and part and parcel with, the flawed approach of the Bulgarian system generally to the issue of lack of consent. It was because of the criterion of force/resistance that the investigation did not go into matters which otherwise it should have done. This must be the explanation for the observations about investigation, for otherwise it is very difficult to see that the criticisms made could found a breach of article 3 given the words of paras 167 and 168. Those paragraphs make it clear that the gloss on article 3 is not a vehicle for the second-guessing via the Convention of the ordinary domestic process of assessment of conflicting evidence. Of course it is true that the Strasbourg court is a supra-national one, but there is no sign that the limit on concern with “errors or isolated omissions” is restricted to that court as distinct from a national court when the latter is applying the Convention; on the contrary, the limit is expressed to be one which determines when there is a breach of article 3 and this plainly is the same for all courts examining that question. Nor can it be the case that a system which does not involve pre-trial confrontation of witnesses, as some Code Napoleon systems do, but other systems, including all the United Kingdom ones, do not, is ipso facto in breach of article 3.

What has happened since MC v Bulgaria is that the formulation cited above has been repeated, or in some cases summarised. It is plain that in several of them the summary has been to the effect that whenever there is an allegation of ill-treatment passing the article 3 threshold, by whomever committed, there is an obligation on the state to conduct “an effective investigation”.

But in none of these cases has the basis for, and thus the ambit of, any obligation to investigate third party violence ever been addressed. Reference back to MC v Bulgaria, and to its reliance on Assenov, Calvelli and Osman is frequently made, but never examined and the uncertainties mentioned above have not been confronted. In some cases, there is additional reliance on a series of Turkish cases: Ay [2005] ECHR 167, Ali & Aysu Duran [2008] ECHR 289, Mehmet [2008] ECHR 269, Celik [2004] ECHR 548 and Bati [2008] ECHR 246. But all of these were cases of allegations of serious police torture or ill-treatment of suspects; the citation of such cases illustrates the manner in which the difference between the first gloss and the second has not received attention.

In most cases the injunction that the court is not concerned with errors or isolated omissions is similarly repeated. In some, the court appears to have looked accordingly for evidence of a structural defect, alternatively “culpable disregard” or an absence of good faith, in the administration of the domestic system: see for example Szula v United Kingdom (2007) 44 EHRR SE19 and BV v Croatia (Application No 38435/13) 21 January 2016. But this is not always the case. The injunction notwithstanding, in some of the cases the criticisms of the investigation have been very particular. In some, there are plain overtones of structural State deficiencies in relation to the investigation of allegations of racially motivated or discriminatory violence: examples include Secic v Croatia (2009) 49 EHRR 18, Beganović v Croatia [2009] ECHR 992, and Milanovic v Serbia [2010] ECHR 2029, but this was not given as a reason for the decisions. In others, such as Vasilyev v Russia [2009] ECHR 2078, the nub of the allegation was serious misbehaviour by the police, in that case by dumping the injured men in the street in the aftermath of a third party assault. But in some, the court has found itself simply pronouncing on whether the investigation was sufficiently careful. An example appears to be CAS v Romania (2015) 61 EHRR 18, where the complaint was of sexual abuse of a boy in his home by a visitor or neighbour. It is possible that the basis of the decision was a conclusion that, as in some other cases, the state did not take allegations of sexual abuse seriously enough, but that is not said. The court found a breach of article 3 not only in delay, which did not prevent the prosecution of the specific individual identified, but also in what it described as a failure sufficiently to evaluate conflicting testimony, supported by the fact that when that individual accused was acquitted, the police did not look for anyone else. That is very close to, if not indistinguishable from, a view that the outcome of the trial was wrong, and moreover that if it was wrong that could itself amount to a breach of article 3. Beganović is another case in which (at para 77) the court appears to have held that the positive obligation under article 3 may extend into examination of the conduct of the trial. Such instances, which may be atypical, leave unanswered the question what is meant by the principle that a breach of article 3 is not constituted by a bona fide decision in the course of investigation or law enforcement which is afterwards held to have been an error.

Occasionally in the Strasbourg cases, the general statement of principles includes the seminal passage from Osman v United Kingdom (para 116) which is set out at para 112 above. In most of them it does not. But it is surely clear that these considerations relating to the practical business of policing, to the operational choices which have to be made as to priorities and to the allocation of finite resources, must apply with equal if not greater force to the investigation of allegations of past third party violence as they do to reports of threats of future violence. These considerations point firmly to the proper test for the ancillary positive obligation under article 3 to investigate reports of past violence being whether the state has a proper structure of legal and policing provision designed to punish it when it occurs and has administered that structure in good faith and with proper regard for the gravity of the behaviour under consideration. They do not point towards a test of ex post facto assessment of whether the investigation was careless or made mistakes which ought not to have been made, nor to a finding that there has been a breach of the right not to suffer torture, or inhuman or degrading treatment, when the complaint is that an investigation could and should have been done better.

This is the more so when one considers the range of behaviour which is treated as triggering the ancillary positive obligation under article 3. It is clear, and regularly emphasised, that treatment must pass “a minimum of severity” before it falls within article 3. It is also clear, and routinely stated, that what that minimum amounts to is relative and varies according to the circumstances of the case, such as the nature and context of the punishment, the manner and method of its execution, its duration, its physical and mental effects, and in some instances the sex age and state of health of the victim. That statement derives from an early corporal punishment case, Costello-Roberts v United Kingdom (1995) 19 EHRR 112, para 30. It was repeated in A v United Kingdom at para 20 and in many subsequent judgments. It is no doubt plain that in the case involving breach of the state’s primary obligation not to inflict inhuman or degrading treatment on its citizens, almost any physical injury perpetrated upon a citizen by a State official, typically the police, will cross the article 3 threshold. Save where necessary in the course of resisted arrest or the prevention of crime, there is simply no place in policing for causing injury of any kind to a suspect. The practice of the Strasbourg court in relation to third party violence might have built upon the relative nature of the threshold in order to limit the ancillary positive obligation to very serious violence but it has not done so. In Milanovic v Serbia numerous cuts combined with feelings of fear and helplessness were regarded (at para 87) as triggering the ancillary positive obligation. In BV v Croatia (at paras 153 and 121) injuries to the head and contusions to the body were treated likewise. It is plain from the cases that the threshold is regarded as falling, in English terms, somewhere on the scale of actual bodily harm. To that must be added rape and child sexual abuse; whether indecent assault passes the threshold remains unclear but it is perhaps likely that it may. So also one would think must be added false imprisonment (for example by relatives), violent disorder, most terrorist offences and many other crimes. It follows that the great majority of violent and sexual offences will trigger the ancillary positive obligation, and that potentially the investigation of all such offences might lead to an action under section 6 of the Human Rights Act, querying the adequacy of the police treatment of the case.

It might also be noted that the application of the judicial glosses to the other rights protected by the Convention has not, as yet, received detailed consideration. But it is difficult to see why, if they are sound, they may not in principle be applied equally to other rights. In Siliadin v France [2005] ECHR 545 the Court held that the second gloss applied to article 4 at least as far as the obligation to put in place legal prohibition of forced labour was concerned, but an obligation in relation to investigation was not in issue. Some third party behaviour in relation to modern slavery might indeed be considerably more serious than actual bodily harm in a fight outside a club. Some of the cases clearly contemplate that third party infringement of article 8 rights may trigger the ancillary positive obligation: see in CAS v Romania at para 72, and Szula v United Kingdom at para 1. The possible application of that ancillary positive obligation to third party interference with the right to enjoyment of one’s possessions under Article 1 Protocol 1 has yet, it seems, to be considered. But in principle, the state has a duty to protect this right in its citizens, as it has in relation to all the other rights under the Convention. If so, the prospect may exist of the response to every complaint of burglary, car theft or fraud becoming the subject of an action under the Human Rights Act.

English law recognises a public legal duty owed by the police to enforce the law. The police enjoy a wide measure of discretion as to how to go about it, what inquiries to make, and when and whom to prosecute, but a structural failure to enforce a particular part of the law is amenable to direction by the court via judicial review on the application of any interested party: R v Comr of Police of the Metropolis, Ex p Blackburn [1968] 2 QB 118. Such a public duty is real, not abstract. That is consistent with the implied positive obligation recognised by the Strasbourg court in cases such as A v United Kingdom. English law also recognises the liability of the police to individuals where a tortious duty of care is broken, as it may be where they have directly or indirectly occasioned physical harm: Knightley v Johns [1982] 1 WLR 349 is a simple example. Further, it has a statutory scheme for the independent investigation of complaints about the police, and a different statutory scheme for the compensation of citizens who are injured through the criminal acts of others. What English law does not recognise is a duty of care in tort owed by the police to individual citizens and sounding in damages in relation to the detection of crime and the enforcement of the law.

The reasons for this absence of private law duty of care were fully explained by the House of Lords in Hill v Chief Constable of West Yorkshire Police [1989] AC 53, and confirmed by that court in Brooks v Comr for the Police of the Metropolis [2005] 1 WLR 1495, [2005] UKHL 24 and in Van Colle v Chief Constable of Hertfordshire Police [2008] UKHL 50, [2009] AC 225; Smith v Chief Constable of Sussex Police [2008] UKHL 50, [2009] AC 225; and by this court in Michael v Chief Constable of South Wales Police [2015] AC 1732, [2015] UKSC 2. They find convincing expression in the first three cases in the separate speeches of Lords Keith, Steyn, Hope, Phillips, Carswell and Brown and were supported also by Lord Bingham despite his solitary dissent on the extent of the particular duty (of prevention) under consideration in Smith. A convenient summary is perhaps afforded by the judgment of Lord Phillips in Smith at para 97:

These reasons are powerful, repeated and carefully considered. They are grounded in public policy and have something in common with the considerations explained by the Strasbourg court in Osman v United Kingdom at para 116 (see above at para 112). In the briefest of terms, law enforcement and the investigation of alleged crime involve a complex series of judgments and discretionary decisions. They concern, amongst many other things, the choice of lines of inquiry, the weighing of evidence thus far assembled and the allocation of limited resources as between competing claims. To re-visit such matters step by step by way of litigation with a view to private compensation would inhibit the robust operation of police work, and divert resources from current inquiries; it would be detrimental to, not a spur to, law enforcement. It is not carrying out the impugned investigation efficiently which is likely to lead to diversion of resources; on the contrary. It is the re-investigation of past investigations in response to litigation which is likely to do so. Moreover, whilst there may exist a mechanism by way of summary judgment for stopping short such a re-investigation if the litigation be “spurious” in the sense of demonstrably bad on the papers, other claims, and particularly those which turn out to be speculative, cannot thus be halted. In short, the public duty would be inhibited by a private duty of such a kind. A contemporary example can be seen in terrorist activity. It is well known that large numbers of possible activists are, to some extent or other, known to the police or security services. The most delicate and difficult decisions have to be made about whom to concentrate upon, whose movements to watch, who to make the subject of potentially intrusive surveillance and so on. It is in no sense in the public interest that, if a terrorist attack should unfortunately occur, litigation should become the forum for a review of the information held about different suspects and of the decisions made as to how they were to be dealt with. Nor is it difficult to see that it is by no means necessarily in the public interest that there should be pressure on the authorities, via the prospect of litigation, to ratchet up the surveillance of additional persons. The long-standing controversy over police use of powers of “stop and search”, for instance in relation to the carrying of knives by youths, affords another example.

It may be noted that in his separate opinion on Osman in the commission, Sir Nicholas Bratza recognised the force of these considerations: (1998) 29 EHRR at 298. As he pointed out, the difficulties highlighted by the House of Lords in Hill were well illustrated by the facts of Osman. The allegations there made by the claimant would, he said, involve an investigation not only into issues of fact but into acutely difficult questions of policy and discretion.

The relevance of the position at which English law has arrived, after long consideration at the highest level, is not that English law can control the operation of the ECHR. But it is to highlight the delicate balance between the duty of the State to the public generally and its relationship to individuals in particular cases of reported or anticipated crime. That delicate balance is as applicable to the ambit of the implied ancillary positive obligation under articles 2, 3 and maybe others as it is to an action in tort in the domestic courts. When taken together with the uncertainties as to the nature and extent of the implied positive obligation as found by the second gloss on those articles, it points clearly to the undesirability of any assertion of a detailed review of the course of a particular criminal investigation by way of the Convention. It is one thing to say that a state must take seriously its protective obligation, must put in place structures which enforce the law and must not then ignore them. It is quite another to say that by way of the Convention every police investigation should be examined in detail to see whether it should have been done better, and that compensation should be paid out of the limited police resources, at the expense of other necessary expenditure on current cases, if the decision is that it should have been. These important public considerations have nowhere been examined or put into the balance in any of the Strasbourg cases on the second gloss, from MC v Bulgaria onwards.

It can properly be said that the distinction between structural and operational defects may at times be difficult to make. It is, however, no more difficult than the distinction which it is suggested must be made, if operational negligence suffices, between errors which amount to breaches of article 3 and those which do not. To say that the errors must be serious and significant in order to amount to a breach of article 3 is surely more to present than to solve the difficulty. Nor is such a restriction clearly to be found anywhere in the line of Strasbourg cases relied upon. If the test is not to be simply a falling below the standard to be expected of the police, and thus the same as negligence, it is not easy to see what it is.

The English cases make a clear distinction between the objectives served by a tortious duty to compensate and a Convention-based duty to uphold the prohibition on inhuman or degrading treatment: see for example Lord Brown in Van Colle at para 138. In substance, the Convention-based duty is not aimed at compensation but at upholding and vindicating minimum human rights standards. It is, substantially, to insist on performance of a public duty. It is now said that this distinction justifies the acceptance of a general duty under article 3 to investigate any report of past behaviour passing the threshold of that article, because such will not impinge on the common law position as firmly established in the cases beginning with Hill. The error in this argument is to seek to have it both ways. One cannot both uphold the distinction and effectively eliminate it by employing a Convention claim to serve substantially the same purpose as an action in tort. That it will do if the wide ambit of the ancillary article 3 duty is accepted, and if compensation routinely follows a finding that a criminal investigation should have been better conducted. True it is that the limitation period differs, but this will not remove the disadvantages to policing which were identified in the English cases. It may be that there is a more relaxed approach to causation in a Convention-based claim, but that if anything only increases the prospect of such a claim becoming a substitute for a claim in tort. There is no doubt some difference of approach to the calculation of compensation, but the present case is a good illustration of the marginal, if not imperceptible, nature of the distinction in outcome - see the judge’s scrupulous quantum judgment at paras 33, 130 and 143. If, on the other hand, the positive duty under article 3 is recognised to conform to the public duty, to put in place structures to outlaw the prohibited behaviour and to operate them in good faith, the Convention-based claim will afford the possibility of some compensation where the English common law rules do not, but will not result in wholesale substitution of the Convention for a duty in tort.

For the claimants Ms Kaufmann QC mounts an elegant argument. It is that the positive obligation to investigate past third party crime for which she contends is simply analogous to the Osman duty to protect from threatened violent crime. Just as the latter arises when there is a real and immediate risk of prescribed behaviour, so she contends the former arises when there is a credible report that the prescribed behaviour has already occurred.

Elegant as the argument is, the two situations are not analogous. There is a clear distinction between protection from an immediately anticipated danger and inquiry into a past event. The carefully limited Osman duty arises because there is an immediate risk of death or serious injury to an identified individual, communicated to the State. By contrast, a crime of violence committed by A against B will only occasionally carry a risk of repetition, whether against B or against others; there is generally no immediate danger to an identified person.

It is no doubt possible to categorise the duty to investigate reports of third party crime as deriving from the state’s duty to protect its citizens, in the same way as does its duty to have in place structures which make such behaviour unlawful. So viewed, the duty to investigate each reported crime can be said to be an application of cases such as A v United Kingdom. But the reality is that there is a marked and vital distinction, even if it is sometimes of degree, between structural failure to outlaw the behaviour and operational failings in the investigation of particular reports.

The judge’s findings as to what went wrong are not disputed, and were in any event largely based on the very critical internal police reports, as well as on that of the Independent Police Complaints Commission. What went wrong involved plain structural errors. The Metropolitan Police had a written policy for recognising and dealing with cases of drug induced rape but it was institutionally treated as mere form and there was no proper training in its application. The complaints made by the claimants were simply not accorded the kind of weight which they demanded because of generic failures to treat them with sufficient care and gravity. Moreover there was pressure internally to write off cases of the kind here encountered. The various detailed failings in the conduct of the inquiry were largely attributable to this flawed structural approach. They included those set out by Lord Kerr at para 51. There is, as explained at the outset, no appeal as to quantum. In those circumstances, this is a case which falls within the ancillary positive duty under article 3, as it ought to be interpreted. It is for that reason that I agree that the appeal in the present case ought to be dismissed.

I have read with benefit the three judgments prepared in this case by Lord Kerr, Lord Neuberger and Lord Hughes. The result is not in doubt, but there is a significant difference between Lord Kerr and Lord Neuberger on the one hand and Lord Hughes on the other regarding the extent to which the Convention rights, as domesticated by the Human Rights Act 1998, should be seen as imposing on the State an operational duty to investigate serious offences the commission of which there is no reason to attribute to state agents. If there is such a duty, then I do not see that it can or should be confined to the victim of the offence in question. Part of its purpose must be not only to punish, but also to deter and to prevent the occurrence of further such offences, and, if a third person suffers foreseeably as a result of a failure properly to investigate, that third person appears to me, potentially at least, to be a victim.

There is much force in Lord Hughes’ analysis and critique in relation to the question whether any such general duty exists. What has happened in the Strasbourg jurisprudence is, unfortunately, not unprecedented. The European Court of Human Rights starts from a solidly rationalised principle, but then extends it to situations to which the rationale does not apply, without overt recognition of the extension, without formulating any fresh rationale and relying on supposed authority which does not actually support the extension. Further, the European Court of Human Rights has not in the present context really focused at any stage on the implications for policing of the general duty which it has suggested. These have been discussed domestically in a number of common law cases, and include the risks of defensive policing and of police priorities being affected by the perceived risk of being sued, as well as the significant financial implications of exposing the police to all those potentially affected by any failure in police investigative work: see eg Michael v Chief Constable of South Wales Police [2015] UKSC 2; [2015] AC 1732, paras 121-122.

In these circumstances, while appreciating the pressures under which the European Court of Human Rights operates, and the difficulties of maintaining coherence and discipline in a court consisting in the first instance of multiple chambers, an approach, careful to identify, rationalise and justify any significant development of principle, would save domestic litigants and courts time, effort and expense.

The starting point is the positive duty on the State under article 3 not to subject anyone “to torture or to inhuman or degrading treatment or punishment”. One (solidly rationalised) principle which the Court has derived from that duty, by way of gloss, is an ancillary positive duty to conduct “an effective official investigation” where “an individual raises an arguable case that he has been seriously ill-treated by the police or other such agents of the State unlawfully and in breach of article 3”: Assenov v Bulgaria (1998) 28 EHRR 652, para 102; the basis for this being that:

This is a coherent gloss, derived from the rationale of article 3. It was repeated in Sevtap Veznedaroglou v Turkey (Application No 32357/96), para 32, cited in Jacobs, While and Overy’s The European Convention on Human Rights (OUP, 2014) where this rationalisation is cited as the basis of any positive investigative duty. The Supreme Court in R (Smith) v Oxfordshire Assistant Deputy Coroner [2010] UKSC 29; [2011] 1 AC 1, expressed a similar understanding of the parallel ancillary investigative duty capable of arising under article 2 (providing that “Everyone’s right to life shall be protected by law”): “the investigative obligation under article 2 arose only in circumstances where there was ground for suspicion that the state might have breached a substantive obligation under article 2”, and “the death of a soldier on active service did not of itself raise a presumption of such a breach”: see headnote, holding (2). At paras 200-212, I examined both the Strasbourg and the domestic jurisprudence on this point, including the analysis of the point by the House of Lords in the prior authority of R (Middleton) v West Somerset Coroner [2004] 2 AC 182, para 3 per Lord Bingham, and R (Gentle) v Prime Minister [2008] AC 1356, para 6, per Lord Bingham. The investigative duty was in short parasitic.

At para 210, I examined the various specific situations in which the European Court of Human Rights had held that there was sufficient State involvement to give rise to a substantive obligation to protect, combined with a parasitic duty to investigate after the event: killings by State agents and deaths of persons in custody or mental health detainees, deaths of conscripts, as well as situations where under the principle in Osman v United Kingdom (1998) 29 EHRR 245, the State was on notice of a specific and immediate threat to someone’s life and bound accordingly to take protective steps.

In para 211, I pointed out the distinction between the procedural investigative obligation arising in such circumstances and “the general substantive obligation under article 2 to establish an appropriate regulatory, investigatory and judicial system”.

To my mind, Lord Hughes’ analysis fits perfectly with what the House and the Supreme Court then understood to be the law under article 2, and I believe would also have thought the law to be under article 3, since there is no reason to differentiate in this respect between them. The case of MC v Bulgaria (2005) 40 EHRR 20 was not even cited in R (Smith) and there is reason to believe that its potential significance under these articles was not appreciated until much more recently: see paras 149 and 150 below.

It is at this point that the European Court of Human Rights extended a solidly rationalised principle to situations to which the rationale did not apply. What it did in MC v Bulgaria (2005) 40 EHRR 20, para 151, was to cite Assenov, para 102, in support of propositions that:

On this authority has however been piled a weight of subsequent Strasbourg caselaw, including some recent Grand Chamber authority, to the effect that the investigative duty is not, or not necessarily, confined to cases of suspected misdoing or default by State agents. Lord Kerr and Lord Neuberger have examined this caselaw. While its foundations or rationale may be shaky, I cannot ignore at any rate the clear terms in which the conclusion has now so often been expressed, to the effect that the State’s positive investigative obligation can arise even where the relevant offence is not arguably attributable to any State agent.

There are however some caveats that I would make, based on such explanations as the Court has given as to the working of this extended duty:

The Court has reiterated that the “scope” of the State’s positive obligations might differ between cases where treatment contrary to article 3 of the Convention has been inflicted through the involvement of State agents and cases where violence in inflicted by private individuals: see eg Beganović v Croatia [2009] ECHR 992, para 62, Vasilyev v Russia [2009] ECHR 2078, para 100 and other cases cited by Lord Neuberger in para 88. This must mean something in practice, even though the Court went on to say that “the requirements as to an official investigation are similar”.

The Court has also repeatedly emphasised that it is not concerned with “allegations of errors or isolated omissions in the investigation”. A statement to that effect appears in the seminal authority of MC v Bulgaria, para 168, and is repeated in CAS v Romania (2015) 61 EHRR 18, para 69, BV v Croatia (Application No 38435/13) and BV v Belgium (Application No 61030/08), paras 55-61. In place of what was once understood to be a distinction between casual errors of judgments or acts of negligence, consisting of “operational” as opposed to systematic failures by state servants or agents, there is now a distinction to be drawn between simple errors or isolated omissions in the investigation and more serious failings. In this connection, I agree with Lord Hughes, para 123, that there is no basis for treating this qualification on the scope of the positive investigative duty under article 3 as confined to Strasbourg as a supra-national court, and as irrelevant to the English domestic courts’ interpretation of article 3. I had understood Lord Kerr to suggest the contrary, but he has clarified in paras 27 to 30 that this is not the case.

In paras 27 to 30, Lord Kerr expresses a conclusion that the only shortcomings relevant when it comes to the operational duty to conduct an investigation are those which are “conspicuous or substantial”, or “really serious”, or “egregious” or “obvious and significant”. Lord Hughes considers that this is “more to present than to solve the difficulty”, and that no such restriction is “clearly to be found anywhere in the line of Strasbourg cases relied on”: para 136. But a distinction between mere shortcomings and more serious failures is at least consistent with the Court’s statements of principle set out in the previous subparagraph, and appears in the reasoning in Mustafa Tunç v Turkey (Application No 24104/05), paras 189, 192 and 195. It is also consistent with the Court’s more general jurisprudence, to the effect that:

The investigative duty currently under consideration should not be confused with, and cannot be treated as part of, an Osman type duty on the state to act in the face of a real and immediate threat imperilling the life or bodily well-being of a potential victim. That would involve mining and extending a separate strand of Strasbourg caselaw. In some cases, it would mean there was no investigative duty at all, if no real and immediate threat was apparent to anyone, or anyone other than the original victim, following the original offence. The investigative duty which the Strasbourg case law, in my opinion, now recognises is not tied down by any such restriction. It arises from the fact of the offence. I endorse what Lord Hughes says on this aspect in his paras 137-138.

Finally, I do not accept that Lord Bingham’s well-known cautionary remarks in R (Ullah) v Special Investigator [2004] UKHL 26; [2004] 2 AC 323 were confined to the international level (whatever relevance that would mean they had domestically). They were, and have correctly been understood in later authority, such as R (Al-Skeini) v Secretary of State for Defence [2007] UKHL 26; [2008] AC 153, as guidance relating to the general approach which domestic courts should take. The general aim of the Human Rights Act was to align domestic law with Strasbourg law. Domestic courts should not normally refuse to follow Strasbourg authority, although circumstances can arise where this is appropriate and a healthy dialogue may then ensue: see eg R v Horncastle [2009] UKSC 14; [2010] 2 AC 373; Manchester City Council v Pinnock [2010] UKSC 45; [2011] 2 AC 104, para 48 and Chester v Secretary of State for Justice [2031] UKSC 63, [2014] 1 AC 271, paras 27-28. Conversely, domestic courts should not, at least by way of interpretation of the Convention rights as they apply domestically, forge ahead, without good reason. That follows, not merely from Ullah, but, as Lord Hoffmann said in In re G (Adoption: Unmarried Couple) [2008] UKHL 38; [2009] 1 AC 173, para 36, from the “ordinary respect” attaching to the European Court of Human Rights and “the general desirability of a uniform interpretation of the Convention in all member states”.

There are however cases where the English courts can and should, as a matter of domestic law, go with confidence beyond existing Strasbourg authority: see eg Rabone v Pennine Care NHS Foundation Trust [2012] UKSC 2; [2012] 2 AC 72. If the existence or otherwise of a Convention right is unclear, then it may be appropriate for domestic courts to make up their minds whether the Convention rights should or should not be understood to embrace it. Further, where the European Court of Human Rights has left a matter to States’ margin of appreciation, then domestic courts have to decide what the domestic position is, what degree of involvement or intervention by a domestic court is appropriate, and what degree of institutional respect to attach to any relevant legislative choice in the particular area: see In re G, paras 30-38, per Lord Hoffmann, para 56, per Lord Hope and paras 128-130, per Lord Mance.