Drowning Migrants, the Human Rights Committee, and Extraterritorial Human Rights Obligations

In this post I will analyse more extensively the two decisions of the UN Human Rights Committee that I flagged previously (A.S. and others v. Malta, CCPR/C/128/D/3043/2017 ; A.S. and others v. Italy, CCPR/C/130/DR/3042/2017), dealing with the failure of Malta and Italy to rescue a group of more than 200 migrants whose vessel sank in the Mediterranean in 2013. My focus here will be on the threshold question of the extraterritorial application of the Covenant on Civil and Political Rights, which revolves around the concept of state jurisdiction in Article 2(1) ICCPR.

Briefly, the Committee applied a functional conception of jurisdiction to find that the migrants were concurrently within the jurisdiction of both Malta and Italy. The case against Malta was however declared inadmissible for failure to pursue any domestic remedies, while Italy was found responsible for failing to discharge its positive obligation to protect the lives of persons in distress at sea. For more background and analysis, see here, here, here and here.

As I explained in my initial post, this was the Committee’s first big opportunity to apply the functional approach it had first articulated in General Comment 36 (2018) on the right to life, and to do so in the most difficult of contexts – the applicability of positive obligations of protection. And while the Committee found that both states had jurisdiction concurrently, it was actually quite divided in doing so. The decision that the authors of the communication were subject to Malta’s jurisdiction was nearly unanimous, 13 votes to 1 (Andreas Zimmermann dissenting), and was grounded primarily in the fact that the shipwreck had happened within Malta’s designated search and rescue area. The decision that the authors were also concurrently within Italy’s jurisdiction was, however, made by 9 votes to 6 (and by a slightly different composition of the members of the Committee, due to turnover and elections). The individual opinions are very much worth the read.

Perhaps the most remarkable thing about this pair of cases is how all of the Committee’s members, in the majority or the minority, are attempting to delineate when positive obligation of protection under the Covenant would apply, and how none of them are able to avoid arbitrary line-drawing. And that’s probably because non-arbitrary line-drawing just can’t be done in a rule-bound, quasi-judicial setting when it comes to positive obligations that effectively require the redistribution of resources and burdens in the context of a collective action problem, such as the rescue of migrants in the Mediterranean, where each option is imperfect and results in trade-offs. The Committee’s members are all concerned with either going too far or not going far enough, even if obviously they disagree on the final outcome. But let’s now look at the two decisions in more detail.

The Malta decision

By 10 votes to 4 the Committee found the communication against Malta inadmissible for failure to exhaust domestic remedies. It could therefore easily have avoided the Article 2(1) ICCPR jurisdiction issue altogether, but it chose not to, i.e. the Committee wanted to send a reasonably clear signal regarding the applicability of the Covenant to persons in distress at sea. Whether that signal is actually clear enough is a different matter.

In para 6.4 of the decision the Committee invoked General Comment 31 (2004), para. 10, which had set out a personal conception of state jurisdiction in the sense of Article 2(1) ICCPR as exercise of ‘power or effective control’ by the state over the victim. In para. 6.5. the Committee then invoked General Comment 36 (2018), para. 63, which employed a more novel, functional conception of jurisdiction as state control over the victim’s enjoyment of the right to life, and which includes persons ‘located outside any territory effectively controlled by the State, whose right to life is nonetheless impacted by its military or other activities in a direct and reasonably foreseeable manner.’ The same paragraph also expressly mentioned that ‘States parties are also required to respect and protect the lives […] of those individuals who find themselves in a situation of distress at sea, in accordance with their international obligations on rescue at sea.’

The Committee proceeds to apply this framework in para. 6.7, holding that it is

undisputed that the vessel in distress was located in the SAR area for which the State party authorities undertook responsibility to provide for overall co-ordination of search and rescue operations, in accordance with section 2.1.9 of the SAR Convention and Regulation 33 of the SOLAS Convention. It further notes that it is undisputed that the State party authorities formally accepted to assume the coordination of the rescue efforts at 2.35 p.m. on the day of the shipwreck. The Committee therefore considers that the State party exercised effective control over the rescue operation, potentially resulting in a direct and reasonably foreseeable causal relationship between the State parties’ acts and omissions and the outcome of the operation. Consequently, the Committee is not precluded by article 1 of the Optional Protocol from considering the present communication. (emphasis mine)

In other words, the Committee finds that the applicants were within Malta’s jurisdiction because (1) they were in distress in Malta’s SAR area and (2) because Malta had answered their distress call and formally undertook coordination of the rescue operation. Malta thus controlled the operation – note how the language is not one of control over the victims, and not even explicitly one of control over their rights, and note also the emphasis on a ‘direct and reasonably foreseeable causal relationship,’ the language previously used in General Comment 36 as some kind of limiting criterion for triggering Covenant obligations. The Committee essentially seems to be saying that Malta could have acted in such a way as to substantially improve or change the situation of the complainants, had committed to do so, and that it therefore should have acted in such a way, i.e. had an ICCPR obligation to do so.

But the Committee’s reasoning is not wholly in this functional mode. In particular, it is unclear whether the Committee would have found the complainants to have been within Malta’s jurisdiction solely on the basis of the fact that they were within Malta’s SAR area, even if Malta had not answered the distress call and had not committed to coordinating the rescue. Put differently, it is unclear whether the Committee considers factual capability to act as the sole factor for finding jurisdiction, or whether it has to be combined with some kind of extraneous legal obligation emanating from other sources of international law, such as the SAR and SOLAS treaties. This becomes a key point in the Italian case, because the shipwreck does not, in fact, take place within Italy’s SAR area even if an Italian ship is the closest to it.

One Committee member – Andreas Zimmermann – dissented from this holding. For him, jurisdiction should be conceptualized as the exercise of state authority and control over the victims, and the mere setting up of a SAR area is not an exercise of jurisdiction in that sense. Nor should, in his view, the applicability of the Covenant depend on or be conflated with state obligations under UNCLOS, the SAR or SOLAS conventions. The existence of jurisdiction in the sense of Article 2(1) ICCPR does not depend on these extraneous parts of international law. Fundamentally he is driven by the policy concern that the issue of such complexity as the rescue of persons in distress at sea should be left to bespoke, specialized regimes of international law and not be subsumed under human rights law. In their concurring opinion, however, Zyberi, Bulkan and Muhumuza defend the Committee’s approach by saying that the protective obligation imposed by the Covenant is one of due diligence and does not impose disproportionate burdens on states, and therefore does not result in disrupting the bespoke specialized rules of international law that so concerns Zimmermann.

Personally I am sympathetic with Zimmermann’s position, in particular with the point that, however properly interpreted, the notion of jurisdiction in human rights treaties cannot depend on the parallel existence of a specialized regime and the designation of SAR areas. The problem with his approach, which (like the one I have advocated for in my book) essentially limits the positive obligation to rescue to situations of territorial control or physical control over the victim, is that the very clarity of the bright-line rules it imposes inescapably leads to some arbitrary results.

Imagine, for example, a shipwreck scenario within the state’s territorial sea, say 11.99 nautical miles from its coast. Zimmermann would likely argue (and so would I) that the Covenant would apply in that scenario, i.e. that the state would have a due diligence obligation to rescue persons in distress in its territorial seas, just as it would have such an obligation for persons in distress on a mountain or in a lake on its territory. But then imagine that exact same scenario occurring 12.01 nautical miles from the state’s coast, on the high seas. It seems entirely arbitrary to say that the first scenario is a human rights problem, in the sense that the ICCPR clearly applies and imposes a due diligence duty to protect life, but that the difference of a mere 0.02 nautical miles in the second scenario suddenly means that the ICCPR – or other treaty or customary IHRL – no longer applies, full stop, and that this scenario has to be left to specialized areas of international law. The only way of justifying this position, I would think, is that on a purely utilitarian calculus its clarity leads to greater good at the cost of occasional arbitrary outcomes, especially if the lack of clarity would disturb the operation of other rules, as Zimmermann argues it would.

The Italy decision

In its views in the case against Italy, the Committee’s analysis of the relevant legal framework follows the Malta decision almost verbatim (paras. 7.4-7.6), but is different in its application to the facts. First, the Committee notes (7.7) that the shipwreck did not take place in Italy’s SAR area, and that therefore the key question was ‘whether the alleged victims could be considered to have been within the power or effective control of the State party, even though the incident took place outside its territory.’ Note how the question is framed as being one of jurisdiction under the personal model as control over the victims, rather than functionally, as control over the victims’ rights. The Committee then takes into account the fact that Italian authorities answered the distress call, an Italian Navy ship was in proximity to the vessel in distress, and how the coordinating Maltese authorities repeatedly asked the Italian Navy to intervene, which they did only several hours into the incident, once the vessel had capsized.

The Committee thus concluded (7.8):

The Committee considers that in the particular circumstances of the case, a special relationship of dependency had been established between the individuals on the vessel in distress and Italy. This relationship comprised of factual elements – in particular, the initial contact made by the vessel in distress with the MRCC, the close proximity of ITS Libra to the vessel in distress and the ongoing involvement of the MRCC in the rescue operation and – as well as relevant legal obligations incurred by Italy under the international law of the sea, including a duty to respond in a reasonable manner to calls of distress pursuant to SOLAS Regulations and a duty to appropriately cooperate with other states undertaking rescue operations pursuant to the International Convention on Maritime Search and Rescue. As a result, the Committee considers that the individuals on the vessel in distress were directly affected by the decisions taken by the Italian authorities in a manner that was reasonably foreseeable in light of the relevant legal obligations of Italy, and that they were thus subject to Italy’s jurisdiction for the purposes of the Covenant, notwithstanding the fact that they were within the Maltese search and rescue region and thus also subject concurrently to the jurisdiction of Malta. (emphasis added)

On the merits, the Committee found that Italy failed to act with due diligence to protect the right to life (8.5):

The Committee notes that the principal responsibility for the rescue operation lies with Malta, since the capsizing occurred in its search and rescue area, and since it undertook in writing responsibility for the search and rescue operation. The Committee however considers that the State party has not provided a clear explanation for what appears to be a failure to promptly respond to the distress call, prior to the assumption of responsibility for the search and rescue operation by the Maltese authorities. It also notes that the State party has not provided any information about measures taken by State party authorities to ascertain that the RCC Malta was informed of the exact location of the vessel in distress and that it was effectively responding to the incident, despite the information about the deteriorating situation and the need for Italian assistance. In addition, the State party failed to explain the delay in dispatching the ITS Libra, which was located only one hour away from the vessel in distress, towards it, even after being formally requested to do so by RCC Malta. Finally, the Committee notes that the State party has not clearly explained or refuted the authors’ claim that intercepted phone calls indicate that the ITS Libra was ordered to sail away from the vessel in distress. In light of these facts, the Committee considers that Italy has failed to show that it has met its due diligence obligations under article 6 (1) of the Covenant.

The jurisdictional approach here is undoubtedly functional – Italy had the capacity to help the vessel in distress, and was thus obliged to do so. This is clearly a very expansive outcome, which the Committee tries to limit by reference to a special relationship of dependency. But that relationship is not really special or practically limitable. Most states would be bound by the same legal framework (SAR and SOLAS treaties). Answering the distress call and the continued involvement of the Italian MRCC in the incident hardly seem like pivotal factors – if the outcome here is the right one, would it really have been any different if the Italian authorities had deliberately ignored the distress call, i.e. had simply refused to answer it? The key factor thus seems to be the proximity of the Italian vessel, because it is that proximity that leads to the physical capability of the Italian authorities to rescue the complainants (or at least to attempt to do so).

In sum, the Committee’s notion of special relationship of dependency is incoherent, and is simply a half-hearted attempt to deflect criticism that their functional approach is overly expansive. This is not to say that dependency is never a legitimate reason for creating positive obligations. Even in systems that are very reluctant to impose such obligations, and certainly do not (as human rights law does) impose them across the board – a relationship of dependency may trigger such an obligation. Compare, for example, the general approach of the US Supreme Court to constitutional rights, which it generally regards as being negative in nature (see, e.g., DeShaney v. Winnebago County, 489 U.S. 189 (1989) (holding that the failure of state authorities to protect a child from abuse by his parent was not a violation of the Constitution), to instances of dependency in which it does impose positive obligations, i.e. liability for omissions (see, e.g., Estelle v. Gamble, 429 U.S. 97 (1976) (holding that state authorities had to provide medical care and clothing, food, and water to individuals in prison even if they had no such obligation towards the general population outside prison)). Or, think of administrative law doctrines such as legitimate expectations, where individuals can expect to reply on promises made to them by state authorities to provide some specific benefit.

In other words, had Italy made some kind of explicit promise to rescue migrants in distress in the Mediterranean that was somehow ‘special’ and went beyond the general obligations pursuant to the SAR or SOLAS Conventions, a promise the migrants relied on when they embarked on their perilous journey, we could talk about some kind of special relationship of dependency. Or, if the migrant’s boat had capsized because it was hit by an Italian vessel, we could talk about a special relationship of dependency that impose a special duty of rescue on Italy specifically. But this was simply not the case here – it’s not that the migrants were dependent on Italy, but that Italy had the power to save them, as would have been the case with any other state that had ships close enough to the migrant’s boat so that they could feasibly help them. 

Zimmermann’s dissent follows his reasoning from the Malta case. Regarding Italy he zeroes in on the proximity issue, and is concerned about disincentivizing states from deploying ships in areas where boats in distress might be to avoid a situation of ‘dependency.’ Again, as a matter of policy, he feels that the collective action problem in rescuing migrants in distress should be dealt with by specialized regimes of international law.

The dissenting opinion of Shany, Heyns and Pazartzis is particularly interesting, on account of both its reasoning and the prior positions of its authors. Shany was one of the principal authors of GC 36 and a long-time advocate of an expansive, functional approach to extraterritoriality; Heyns similarly endorsed expansive approaches to extraterritoriality during his tenure as UN Special Rapporteur on extrajudicial executions. Yet the broad approach to extraterritoriality endorsed by the Committee majority goes too far even for them. How so?

First, the three dissenters distinguish between the state actually exercising control over individuals, and merely having the potential to do so. But while that distinction is easy to draw for negative obligations (e.g. killing a person v. merely having the ability to do so), because they involve state actions, it is much more difficult to draw for positive obligations, i.e. for situations in which the issue is the state’s responsibility for an omission. The issue in such cases is always a normative one – should the state be obliged to act, assuming it has the ability to do so? And if one is grounding the state’s duty to act in its factual ability to do so, as functional approaches to jurisdiction generally do, then the distinction between the actual exercise of control (over the victim or over their rights) and the mere potential to do so becomes meaningless. If a child is drowning in a pond in front of me and I have the ability to save her, is the child actually under my control or power or only potentially so? This is a distinction without a difference.

Second, at para. 4 of their opinion Shany, Heyns and Pazartzis argue that the shipwrecked vessel came under Italian control once the Navy ship ‘arrived at the scene.’ The three dissenting members thus concede that if a ship of state A is ‘at the scene’ of a shipwreck, then the victims would be within A’s jurisdiction even if the shipwreck is not within A’s SAR area. But the ‘at the scene’ criterion appears to be entirely nebulous, especially when – as in this case – the state’s authorities can choose to bring their ship closer to or farther from the vessel in distress.

Consider, in that regard, the hypothetical in my initial post on the case:

In the middle of the Atlantic, a luxury yacht carrying some really, really nice people (e.g. Mark Zuckerberg, Elon Musk and Jeff Bezos) capsizes. Immediately next to it – like 50 meters away – is an Italian military cruiser. Italian sailors can see before their very eyes how the persons aboard the yacht are in immediate distress. Zuckerberg is holding to a makeshift raft, but his grip is slipping; the waters are very cold.

The capsizing is obviously not taking place within the Italian SAR. Looking at the views of the Committee and the opinions of its members, do you think that the drowning Zuckerberg et al would be subject to Italy’s jurisdiction in this scenario? Would your answer change if the Italian cruiser was a kilometre away? Would your answer change if it was an hour’s sailing away (which was the situation in the HRC Malta/Italy case)? Would your answer change if there were other vessels in the area, some of which were closer? Would your answer depend on whether the Italian cruiser responded, or not, to any distress call?

If, as Shany, Heyns and Pazartzis argue, the positive duty to protect life and rescue those in distress would be triggered ‘at the scene,’ that must cover the Italian cruiser when the other vessel is capsizing just next to it. But surely that must equally be the case if the cruiser is a 100 metres away, or 500 metres, or 1, 2, or 5 kilometres, and so forth – so long as the cruiser is close enough for it to mount an effective rescue effort in the time available. In all of these situations, the fundamental functional question on the merits – did Italy do everything it feasibly or practically could have done to save those in distress – is equally and easily answerable, as it was indeed answerable on the facts of this case. There is therefore no way of drawing a non-arbitrary ‘at the scene’ cut-off line which somehow triggers the applicability of the Covenant. If a genuinely functional approach is taken, it must apply in all situations in which the duty of protection can feasibly be complied with.

Conclusion

I can think of no better example than this pair of cases to illustrate the advantages and disadvantages of a functional approach to the extraterritorial application of human rights treaties. On the one hand it seems to dispense with arbitrary line-drawing. But, especially as applied to positive obligations, it is also so expansive that even its proponents ultimately give up and retreat into arbitrary line-drawing, in order to avoid committing to outcomes that politically or practically seem extreme or infeasible.

This is especially the case in situations, as here, that pose a collective action problem and require burdens to be shared by many states. The universalist moral logic of human rights is ill-suited for such a task. It demands of states in the position to do so to help all whom they can protect, but it can’t decide how these burdens will be distributed among them in a whole series of similar events. Just as in the example of a single child drowning in a pond we wouldn’t say that our individual responsibility to help is lessened by the fact that other people are in a position to help but choose not to, so would the duty of state such as Italy to do all it can not be lessened by the fact that many other states are in a position to help, but choose not to. Yet, on the other hand, the political reality is that such burdens need to be shared for rescue efforts to be effective in the long run.

We could observe the exact same problem in the decision of the Committee of the Rights of the Child on the repatriation of children of foreign terrorist fighters who currently find themselves in horrible conditions in camps in Syria, which applied a functional approach but unduly focused on nationality as a burden-sharing device. And we can observe that problem in similar burden-sharing situations, such as the mitigation of climate change or the redistribution of Covid-19 vaccines from wealthy countries to poorer ones (see here for some relevant questions, and here for one possible set of answers). Similar issues have of course also frequently arisen in domestic law with regard to the imposition of liability for omissions in tort and criminal law, including in particular with regard to any duty to rescue persons in distress, on which there is a wide degree of divergence comparatively, but the burden-sharing aspect is not as crucial there.

It is manifest that policy concerns about coordination and burden-sharing drove all of the Committee’s members – be it the majority ‘special relationship of dependency’; or Zimmermann’s bright-line rules that such issues should be left to dedicated special regimes; or Shany, Heyns and Pazartzis’ attempt to cover only those situations at which the rescue vessel is ‘at the scene’ (see similarly Moore’s dissenting opinion), with all those in the minority feeling that the Committee’s approach, even if motivated by the best of intentions, in the long run might do more harm than good. I have no idea whether that’s right or not, but it is on that judgment, intuitive or empirical, that the case turned. And it is that judgment that led all of the members of the Committee, in the majority in the minority, to avoid espousing a simple, clear articulation of the functional approach – that a state would have the duty to protect life whenever it was in fact capable of complying with that duty.