PARTLY DISSENTING OPINION OF JUDGE PAVLI

1. I have voted in support of the unanimous holdings that Article 14, taken in conjunction with Article 8 of the Convention, is applicable in this case; and that there has been a procedural violation of that provision on account of the flawed investigation into the applicant’s allegations of racial discrimination. I also wish to recognise the ground-breaking nature of this judgment as, together with the judgment in Muhammad v. Spain (no. 34085/17, 18 October 2022, not final) adopted on the same day, these are the first cases in which the Court has considered allegations of racial profiling in police identity checks in a public space.

2. I am writing separately, however, as I do not concur with the majority’s summary conclusion in the last sentence of paragraph 38 of the judgment that, owing to the respondent State’s failure to conduct an effective investigation, “the Court is unable to find” whether there has been a substantive violation of Article 14. As the judgment includes a single operative provision (the second) on the merits of the Article 14 claims, without specifying whether the violation found is of a procedural or a substantive nature, I have been unable to formally vote against the effective finding of no substantive violation of that provision, this being the object of my partial dissent.

3. The Court has concluded under its admissibility analysis that the applicant put forward, at both the national level and in the Strasbourg proceedings, “an arguable claim that [he] had been targeted on account of specific physical [1] or ethnic characteristics”. It reached this conclusion by relying on the uncontested allegation that the applicant and his daughter, being of dark skin, were the only individuals who had been subjected to the identity check in their part of the train; and that the police officer who had performed the check “had not disclosed any other objective grounds for targeting the applicant” (see paragraph 26 of the judgment).

4. This conclusion begs the following question: in the presence of an arguable claim of direct discrimination on racial or ethnic grounds by a State agent, why did the majority not shift the burden onto the respondent to prove that the differentiated treatment was in fact in compliance with Article 14? I find that the judgment provides no persuasive answer to this question in effectively dismissing the applicant’s claim that there has been a substantive violation of the anti-discrimination provision, in addition to the procedural violation. After all, it is a central tenet of our Article 14 jurisprudence that, as a rule, it is for the applicant to show a difference in treatment and for the Government to show that it was justified (see Timishev v. Russia , nos. 55762/00 and 55974/00, § 57, 13 December 2015; D.H. and Others v. the Czech Republic , no. 57325/00, § 177, 13 November 2007; and Di Trizio v. Switzerland , no. 7186/09, § 84, 2 February 2016). It is the very function and purpose of rules on the allocation of the burden of proof to allow the Court to reach substantive conclusions in the absence of complete certainty about the facts of the case or other relevant considerations. Even the respondent Government have conceded that “only the State had the ability to establish the relevant facts” (see paragraph 30 of the judgment). It is therefore not necessary or appropriate to regard the investigative failures at the national level as a factor that would objectively prevent the Court from reaching conclusions on the substantive component. Among other reasons, this may provide perverse incentives to any national authorities which may not be inclined to “lift the lid” on either isolated or, worse still, systemic incidents of racial profiling by State agents. It also makes it nigh impossible for victims of racial profiling to succeed in a claim of a substantive violation in such circumstances.

5. In addition to our own jurisprudence, principles on the reversal of the burden of proof in the discrimination context are also firmly established in European law more generally, including European Union legislation and the standards of the Council of Europe’s own European Commission against Racism and Intolerance (ECRI). These standards are cited in paragraphs 15 ‑ 16 of the judgment, but it is not clear to what purpose. They state the relevant requirements in almost identical terms: when individuals establish “ facts from which it may be presumed that there has been direct or indirect discrimination , it shall be for the respondent to prove that there has been no breach of the principle of equal treatment” (see Article 8 of EU Directive 2000/43/EC, emphasis added; see also ECRI’s General Policy Recommendation No. 7, paragraph 11).

6. Our jurisprudence to date has for the most part looked at conduct of State agents potentially motivated by racist or other discriminatory animus in the context of Article 3 of the Convention, involving, for example, acts of police brutality (see paragraph 33 of the judgment and the cases cited therein). It is true that in some of these cases the Court has chosen not to shift the burden of proof onto the respondents despite the failure of the national authorities to carry out an effective investigation into the allegations of discrimination – on the basis that such an approach would amount to requiring the respondent Government to prove the absence of a particular subjective attitude (see Nachova and Others v. Bulgaria , nos. 43577/98 and 43579/98, § 157, 6 July 2005).

7. The present case, however, sits in a very different context. Firstly, unlike unjustified police violence, which is illegal and can be motivated by many different factors, an identity check of train passengers is presumably legal and ought to be based on sound and objective law-enforcement standards. Secondly, discrimination in this context may not be driven necessarily by a police officer’s individual and conscious attitude or hostility against a particular racial or ethnic group; it may also be the result of biased (or at least permissive) internal police guidelines, practices or attitudes, whether formalised or merely tolerated by the hierarchy. To put it simply, it is not too much to ask of the Government in this context to merely show that the identity check had an objective and reasonable basis, not triggered exclusively or primarily by the person’s race or the fact of belonging to another group. For similar reasons, it would place an unfair and often impossible burden on the applicants to require them to prove the State agent’s discriminatory attitude. This would limit their chances to situations where a police officer would, say, be reckless enough to express his or her discriminatory motives and the future applicant would be lucky enough to have witnesses available. As such, it would hardly be fit to deter pernicious practices of racial discrimination by State agents.

8. Finally, the refusal to shift the burden of proof in cases of an arguable claim of direct discrimination by State agents would create some rather paradoxical effects – considering that the Court has often agreed to do so in situations where applicants have put forward a presumption of indirect discrimination, e.g. by providing evidence of an apparently neutral practice that has produced disproportionately harmful effects on a particular group of people (see Hoogendijk v. the Netherlands (dec.), no. 58641/00, 6 January 2005; D.H. and Others , cited above; and Di Trizio , cited above). In such circumstances, the Government are invited to rebut that presumption by pointing to objective factors underlying the practice or policy. I cannot see why applicants such as Mr Basu, claiming to be victims of direct discrimination in a police check, ought to be placed in a less favourable position.

9. Various international bodies have published findings regarding the degree of prevalence of problematic profiling practices by German law enforcement. The present judgment does not, however, include any information about the general national context. ECRI has addressed the issue in its last two reports on Germany; in the most recent one, from December 2019, it expressed concerns about allegations of racially motivated conduct by police forces and referred, for example, to a study in which 34 percent of respondents of sub-Saharan African background reported having been stopped by the police at least once within the past five years [2] .

10. Another ECRI critique is more directly relevant to this case as it involved the same legal provision that was invoked as the basis for Mr Basu’s identity check in the present case – section 23(1)(3) of the Federal Police Act (the “FPA”, see paragraph 10 of the judgment) – as well as the internal police guidelines on identity checks, which are not publicly available. With respect to the latter, ECRI noted in its above-cited 2019 report (paragraph 107) that “even though a [national] higher administrative court considered the practical guidelines of the police as too vague to protect individuals against their abusive use, ECRI did not receive any information about any attempt to render them more precise”. These were presumably the same laws and guidelines that were in force at the time when the present applicant was stopped during his train ride, on 26 July 2012.

11. German courts have also been critical of section 23 of the FPA and its application in certain contexts. The Baden-Württemberg Higher Administrative Court held in 2018 that section 23 did not provide a sufficient legal basis for an identity check involving a train passenger close to a border crossing; whereas a second court found that the police had misused their powers when carrying out an identity check at a train station where skin colour had been the decisive factor for the police officer in choosing to check the individual concerned [3] . Finally, section 23 of the German FPA has also been subject to review by the Court of Justice of the European Union in a 2017 case involving an individual who had been ID checked by the German police while crossing on foot the Europa bridge between Strasbourg and Kehl. The CJEU found that section 23 was incompatible with the Schengen Borders Code (on internal Schengen checks), considering that the checks were authorised irrespective of the behaviour of the person concerned and with no limitations as to the intensity and frequency of checks [4] .

12. To be clear, these rulings did not hold that German legislation or secondary regulations directly authorised racial profiling by the police. They do strongly suggest, however, that they may facilitate, or are poorly crafted to prevent or deter, such practices by granting too much discretion to the police in making so-called randomised stops and too little objective guidance or restrictions against profiling based on racial or ethnic characteristics.

13. In view of the above submissions, it is my position that the Court should have proceeded to consider whether there had been a substantive violation of Article 14, read in conjunction with Article 8, by assessing whether the respondent Government had been able to rebut the presumption that the applicant had been subjected to discriminatory treatment due to his skin colour. Before turning to the Government’s arguments in this regard, it is necessary to make two preliminary points.

14. The first positive obligation of a State Party in this context is to establish a legislative and regulatory framework that is capable of effectively preventing and deterring police profiling on racial or other prohibited grounds (see, mutatis mutandis , on positive obligations under Article 14, Volodina v. Russia , no. 41261/17, 9 July 2019; Budinova and Chaprazov v. Bulgaria , no. 12567/13, 16 February 2021; and Behar and Gutman v. Bulgaria , no. 29335/13, 16 February 2021) [5] . The judgment’s failure to address these questions is a significant omission in my opinion. In view of the international and national criticism discussed in the second part of this separate opinion, it is rather questionable whether the German legal framework on police checks can be considered compatible with the positive requirements of Article 14. Only a legal system that takes seriously the pernicious effects of racial discrimination, including in the form of improper police profiling of individuals, can be deemed to be in compliance with Article 14. Furthermore, the latest ECRI report includes no information as to whether German anti ‑ discrimination legislation has properly incorporated the EU and ECRI standards on the distribution or reversal of the burden of proof in this field.

15. The next preliminary question relates to the substantive standard to be applied in this context – in other words, what exactly is it that Article 14 prohibits when it comes to profiling by State agents? This question has also remained without a clear answer, as the majority declined to address the claim of a substantive violation. I would argue that to some extent this is true also for the twin case of Muhammad v. Spain (cited above), which did consider the allegations of a substantive violation on the merits. The Muhammad judgment phrases the relevant question as to whether the police were “motivated by animosity against citizens who shared the applicant’s ethnicity” (ibid., paragraph 100) or “motivated by racism” (ibid., paragraph 101) – thus placing a strong emphasis on the police officer’s subjective attitude or animus as the sole basis for a finding of discriminatory treatment. In my view, the approach adopted by the UN Human Rights Committee in Rosalind Williams Lecraft v. Spain (CCPR/C/96/D/1493/2006) is more appropriate in this context, the key question being whether the complaining individual “was singled out ... solely on the ground of her racial characteristics and that these characteristics were the decisive factor in her being suspected of unlawful conduct” (see paragraph 11 of the present judgment, citing paragraph 7.4 of the HRC Views).

16. Turning now to the facts of the present case, I have already noted the unanimous finding that the applicant submitted an arguable claim that he had been singled out on the basis of his skin colour – relying on facts from which an instance of direct discrimination by a State agent can be presumed. To rebut this presumption the respondent Government put forward two main lines of argument: (i) that the applicant and his daughter were not the only persons checked on the same train; and (ii) that an internal police investigation did not find any evidence of racist attitudes on the part of the police officer who conducted the search (see paragraph 30 of the judgment).

17. I find that neither of these arguments, taken alone or together, are capable of meeting the respondent’s burden of proof. The fact that other passengers were also checked, perhaps in other parts of the train, does not prove much in the absence of any data on the racial or ethnic affiliation or appearance of those other checked passengers, or the reasons for checking their identity. Likewise, the previous history of the police officer may be of some relevance, but not decisive on its own. The respondent Government’s arguments still leave us in the dark regarding two crucial considerations: (i) on what basis did the police officer make the specific decision to conduct an identity check of the applicant and his daughter on that train; and (ii) on what basis are such checks generally conducted by German border police under section 23(1)(3) of the FPA. These questions were not answered either at the national level or before the Court. To merely argue that the checks were randomised does not answer the question (and I return to the randomisation aspect below), especially considering the existence of internal guidelines of the German police, whose content remains unknown to the Court, on these issues.

18. The respondent Government have therefore been unable to rebut the presumption of direct discrimination on grounds of skin colour, by pointing to any objective, reasonable and colour-blind grounds for the differentiated treatment. If the Government’s burden was made heavier due to the shortcomings of the investigation by the national authorities, I see no reason why this should work to the Government’s favour or to the applicant’s disadvantage (he would otherwise suffer twice from the poor domestic investigation). As a result, I would have found that there had been a substantive violation of the applicant’s rights under Article 14, read together with Article 8, of the Convention.

19. Today’s cases have only begun to scratch the surface of the complex legal and policy questions surrounding (potentially) discriminatory profiling in Europe. By adopting a highly proceduralist approach and opting not to engage with these difficult questions of substance, the majority have done no favours to the cause of equality of individuals, the development of our jurisprudence, or even the goal of better policing around the continent at a time of great geopolitical turmoil and cross-border challenges.

20. Without seeking in any way to provide an exhaustive analysis or complete answers, the following key questions can be identified:

(i) It has been argued that in order to prevent discriminatory profiling by the police or other related encroachments on individual liberty, national laws should require that identity checks be carried out only on the basis of a reasonable suspicion of illegality (see, for example, the submissions of the applicant in Muhammad v. Spain , cited above), or at least on objective grounds related to the conduct of the individual being stopped and/or police intelligence pointing to such grounds (including physical characteristics of a suspect, e.g. as identified by witnesses to a crime). This option would allow little or no room for so-called randomised checks or checks that are not based on individualised suspicion.

(ii) Conversely, should randomised or sample-based identity checks be permissible in some contexts, such as border controls, mass crowd events, anti-terrorism preventive operations or other situations involving a large number of people? If so, how can this carve-out be prevented from turning into a legal loophole that grants police officers the power to stop people at a whim, whether in law or in practice? If computer programs can perhaps be truly random, human beings are less likely to be so in the absence of previously agreed randomisation methods (e.g. to check every fifth car) that limit the discretion of individual officers. It is worth noting that in the present case the Government’s claims of a random checking of the applicant and his daughter appear to fall in the former category, at least in the absence of any information about the internal guidelines followed by German border police.

(iii) And is there room for a standard that falls somewhere in between reasonable suspicion and randomised checks? For example, some domestic courts have taken the position that identity checks based on a person’s racial or ethnic appearance – however reliable those assumptions might be in the first place [6] – are generally not permissible, unless the police meet the higher burden of demonstrating, through reliable statistics, “an increased delinquency of certain target [racially or ethnically defined] groups on the basis of situational pictures related to the location or situation” [7] . From the perspective of the individual being stopped, would such a standard be considered a reasonable law enforcement policy or one that essentially legalises discrimination by affiliation or association? (On the latter topic, see Molla Sali v. Greece [GC], no. 20452/14, 19 December 2018; and Škorjanec v. Croatia , no. 25536/14, § 55, 28 March 2017).

(iv) As a legal and practical challenge, in many countries the collection of statistics or operational police records on racial or ethnic grounds is prohibited by anti-discrimination law. Paradoxically, however, that may make it much harder for both public bodies and potential victims of profiling to assess and demonstrate the possible existence of indirect discrimination through entrenched or informal practices in law enforcement and other domains [8] . Likewise, internal police guidelines on identity checks are public in some countries (such as Spain), but not in others (like Germany).

(v) Finally, as a broader policy question, does racial or ethnic profiling make for good policing? There is a great deal of research that answers that question in the negative, considering that discriminatory profiling can be seen as an “easy” and ineffective substitute for sound crime-fighting methods; and not least because it tends to alienate entire communities whose cooperation with the police is ever more important in our increasingly multi-ethnic societies [9] .

21. It is of course not possible, or wise, for one or two ground-breaking judgments of this Court to seek to address all facets of a complex phenomenon such as discriminatory profiling in policing. But one has to start somewhere. The facts of the present case, as well as the applicant’s specific submissions, invited the Court to begin to delineate the substantive standards to be applied in this field, beyond the preliminary (albeit essential) requirements of an effective domestic investigation. The majority have declined that invitation by stopping at a finding of a procedural violation. While minimalism may have its fans, both as a legal doctrine and as a school of architecture and design, it is not necessarily the best way to ensure equality for all in our diverse societies; which are here to stay.

[1] I take “physical characteristics” in this sense to mean racial or similar features of a person’s appearance; conversely, one may have physical characteristics (such as blue eyes or being very tall) that do not give rise to discrimination on any prohibited ground.

[2] ECRI, Report on Germany (sixth monitoring cycle), 10 December 2019 (published on 17 March 2020), paragraph 104; referencing EU Fundamental Rights Agency, Second European Union Minorities and Discrimination Survey, 5 December 2017, p. 69.

[3] See, respectively, Higher Administrative Court Baden-Württemberg, 1 S 1469/17, 13 February 2018; and Higher Administrative Court Nordrhein-Westfalen, 5 A 294/16, 7 August 2018, § 74-75.

[4] Court of Justice of the European Union, C-9/16, 21 June 2017 (EU:C:2017:483).

[5] See also, in this respect, Muhammad v. Spain (cited above), Dissenting Opinion of Judge Krenc, §§ 9-13.

[6] For example, the Spanish police collects statistics and categorises identity checks carried out at police stations on the basis of someone’s (apparent) provenance from a certain continent. How does that account for the sometimes widely varying appearance of people coming from the same continent?

[7] Higher Administrative Court Nordrhein-Westfalen, cited above, § 73 (unofficial translation).

[8] See on this point the ECRI recommendations in the most recent country report on Germany (cited above, paragraph 108).

[9] See among others, EU Fundamental Rights Agency, Towards More Effective Policing: Understanding and Preventing Discriminatory Ethnic Profiling , October 2010, pp. 33-44; Council of Europe Parliamentary Assembly, “Ethnic profiling in Europe: a matter of great concern” (Resolution 2364, adopted on 28 January 2021), Explanatory memorandum by Mr Cilevičs, Rapporteur, pp. 10-11; UN High Commissioner for Human Rights, “Preventing and countering racial profiling of people of African descent: Good Practices and Challenges”, January 2019, pp. 9-10; and Open Society Justice Initiative, Profiling Minorities: A Study of Stop-and-Search Practices in Paris , June 2009.