As I explained in my quick-reaction post, the principal disagreement among the judges of the ICJ in Gambia v. Myanmar (GvM) – on which the Court split 9 to 6 – related to proving the existence of genocidal intent. Some of that disagreement was on matters of principle; some of it was on the application of principles that all judges agreed on to the facts. This is not unusual in any judicial context. Some of that disagreement also related to prior definitional questions of what genocidal intent actually is, especially in the context of state responsibility.
The GvM judgment therefore poses two key questions, one descriptive and one normative: did the Court change its approach to defining or proving genocidal intent compared to its prior judgments in the Bosnian and Croatian genocide cases? And, if it did so, was this change justified?
These are, I think, the two most important questions in GvM, questions which largely determined the outcome of the case. These are also the questions that have the greatest implications for any future cases. There are, of course, many others, at various levels of specificity. Here, however, I will only deal with the two questions set out above.
The ‘change’ framing of my first, descriptive question obviously presupposes an understanding of what change means in this context. By ‘change’ here I will mean a genuine and substantial departure from a position that the Court has taken previously; that departure can be explicit or implicit. The key point, though, is that the change is one capable of affecting outcomes, rather than simply being a matter of terminology, precision or tinkering around the edges. The change must also relate to a matter on which a prior position was taken. By contrast to change so defined, we would have the gradual, normal clarification of outstanding issues that were either left open previously or arose here for the first time.
Bearing this definition of what ‘change’ is in mind, and having now digested the judgment and the separate opinions, my own assessment is as follows:
- There is one clear departure from the Court’s prior jurisprudence, which relates to the duty to prevent genocide.
- The Court has not, however, interpreted the definition of genocidal intent differently from prior cases, although there is a bit of interpretative loosening here.
- The Court has remained committed to a high standard of proof (fully convincing evidence) and to using a strict ‘only reasonable inference’ test to prove genocidal intent from circumstantial evidence.
- The Court has usefully clarified some aspects of that test that were not as clear as they could have been in prior jurisprudence, and has also made other important contributions (e.g. with regard to the relationship between genocide and violations of IHL). Most importantly, the Court has rightly said that the ‘only reasonable inference’ test for genocidal intent does not mean that the offender has to act only with such intent, i.e. the commission of genocide can be committed with multiple, compatible intents, one of which is the genocidal one. (Scholars have generally rightly taken this view, although it is frankly astonishing to see just how much confusion there has been on this matter.)
- However, I think it is fair to say that the Court’s overall assessment of the evidence is substantially more flexible than it was in the past. This is the basis for the joint dissent’s criticism that the majority formally maintained fidelity with prior jurisprudence, but has undermined it in practice in how it applied it to these specific facts. There is, in my view, some validity to this criticism as a purely descriptive matter.
- That is separate from the normative point – whether this is what the Court should have done. There are numerous reasons both for and against this evidentiary flexibility. Even more importantly, behind all these reasons is a set of normative, even ideological priors, that inevitably affect all of the judges and each one of us. Bearing that in mind, I think that the majority’s approach was legitimate on the whole; whether this is what I personally would have done is a different matter.
Change as an inevitability
That said, it is important to understand why (a degree of) change in approach was inevitable in this case. First, GvM raised some questions that simply required clarification. There are other pending or possible cases as well, and no one is oblivious to that.
Second, the Court’s changing composition, the importance of which simply cannot be understated. Of the 15 judges (including judges ad hoc) on the GvM bench, only four sat on the 2015 Croatia v. Serbia merits bench (Tomka (as President), Sebutinde, Xue and Abraham), while only two also sat on the 2007 Bosnia v. Serbia merits bench (Tomka and Abraham). It is precisely these four judges who, together with Judge Nolte and Judge ad hoc Kress, dissented in GvM – circumstantial evidence (!) that the GvM majority departed at least to some extent from the majority of the past. Obviously, the lesser the overlap in the composition of the bench, the lesser the imperative to maintain consistency of outcomes.
Third, in Bosnia and Croatia the Court’s basic fact-finding approach, for all the nice stuff it had said about standards and methods of proof and such, was a copy/paste job of the findings of the International Criminal Tribunal for the Former Yugoslavia. The alignment was complete – genocide ‘only’ in Srebrenica, and nowhere else. This was, for various reasons, the right approach. I argued for it 20 years ago (God help me) in my EJIL piece on state responsibility for genocide (at 595-6), before the Bosnia case was decided, and I remain convinced that this was right. There is no ‘civil’ tort of genocide different from the crime of genocide, and the ICTY was far better equipped to do rigorous fact-finding, including as to genocidal intent, than the ICJ ever could.
But in GvM this was simply not an option. There was no ICTY to copy/paste from. The IIFFM and IIMM fact-finding/investigative mechanisms produced valuable reports and collected important evidence, but they were nowhere close to a criminal trial lasting hundreds of trial days, with hundreds of witnesses and thousands of exhibits. Obviously, it was clear to everyone that the ICJ would rely on IIFFMM/IIMM reports, but it was also clear to everyone that they couldn’t be treated with the same degree of authority and reliability as an ICTY judgment (see, in particular, the work of Mike Becker on this). Nor could anything similar be expected in any future genocide case. So unless the ICJ was to throw its hands up and say that genocide could just never be proven before it without an underlying criminal trial, the Court was inevitably going to have to do something different on fact-finding – and that ‘something different’ could not be a substitute quasi-criminal trial lasting many months or even years, because the Court neither has the resources nor the time to do anything remotely like that.
In short, (some) change was inevitable. The question was exactly where and exactly how much – and now we know. We can also be sure that it is not ideal for a Court to be split so heavily, 9 to 6, on a case of this importance – but both compromise and disagreement are also perfectly normal in a court such as this one.
Explicit change, attempted or successful
Some judges in the majority clearly wanted more radical change, but could not persuade their colleagues that this was the right thing to do. Two examples are obvious. First, Judge Tladi, who opens his declaration by arguing that even though in cases brought under Article IX of the Genocide Convention the Court only has subject-matter jurisdiction to decide whether genocide has been committed, and not any other international crimes, in its analysis (but not in the dispositif) the Court should classify atrocities below the genocide threshold ‘as war crimes and/or crimes against humanity if that classification is appropriate, as is undoubtedly the case’ in GvM (para. 3). That is, he expressly wants the Court to make findings that are outside its jurisdiction, but that excess of jurisdiction is in his view avoided by not making these determinations in the operative paragraph of the judgment.
His desire for change is evident also in how he argues that the ICJ’s Bosnia and Croatia judgments are not in any formally binding (which is obviously true) – but the point he is really making is that he is not terribly keen on following what the Court did there (paras. 8-21). That brings us to his main point, which is a proposal to rather drastically depart from the prior jurisprudence of the ICJ and international criminal tribunals alike by reading the specific intent requirement in the definition of genocide so broadly as to include dolus eventualis.
Thus, under his proposal, a person would have genocidal intent not only if he committed the actus reus of genocide with the purpose/direct intent of causing the destruction of a group, but also ‘when the perpetrator knows that the commission of the actus rei in Article II is likely to result in the destruction of the group, accepts that eventuality and nevertheless continues with the commission of such acts’ (para. 23).
The second example I would discuss is that of Judge Charlesworth. After discussing whether genocidal intent in the state responsibility context meant something different than in the individual criminal responsibility context (paras. 2-10), she expressly argued that the Court’s elevated ‘fully conclusive’ evidence standard for proving genocide ‘does not withstand closer scrutiny’ (para. 11) and should be abandoned, even though, in her view, the existence of genocide was proven to such a standard in this case (para. 12). According to the Judge, the Court is essentially dealing with civil liability, and an elevated standard, whether that of fully conclusive evidence or beyond a reasonable doubt, is simply not appropriate (paras. 19-21). Obviously, I disagree – see my ‘no civil genocide’ point above, which Judge Charlesworth generously notes in her opinion – but of course this is a perfectly tenable position. I am a bit confused, however, by her insistence that the ‘only reasonable inference’ standard for inferring genocidal intent would still apply (para. 17), when in fact that standard is wholly a consequence of an elevated standard of proof (see more here).
Now, these two attempts at reform were unsuccessful – neither judge persuaded their colleagues to make this change. But both judges ended up in the majority. And even if they followed the mainstream approach to the definition of genocidal intent or the applicable standard of proof, as part of the Court’s majority, it is difficult to see how their appreciation of the evidence could avoid being on the more flexible side of the spectrum, if I can put it that way. That is only human nature.
The one attempt at radical change that did succeed in persuading the Court relates to the positive duty to prevent genocide under Article I of the Genocide Convention. In the Bosnia case, the Court ruled that this obligation is (1) triggered once a state becomes aware of a serious risk of genocide, but (2) can only be violated if genocide actually occurs (Bosnia, paras. 430-1, reaffirmed in Croatia, para. 441). Now, however, the Court took the opposite view – that the duty can be violated the moment the state fails to act the way it should, i.e. exercise due diligence, even if no genocide is ultimately committed (‘international responsibility for breach of the duty to prevent genocide is not conditioned on a finding that genocide has occurred. Such an interpretation serves the protective humanizing and civilizing purpose of the Convention.’ GvM, para. 487).
This position was supported not only by the ‘main’ majority of 9 judges, but also by Judge ad hoc Kress, who wrote together with Judge Cleveland (who was in the majority) to express his agreement with the point in principle. He did not, however, vote for the part of the operative paragraph in which the Court found that Myanmar failed to prevent genocide by persons not members of its security forces because, in his view, no serious risk of genocide by these persons ever arose – his sep. op., paras 142-9. The joint separate opinion is quite thoughtful; it is interesting not only in terms of substance, but also in having judges across the majority-minority divide writing together on a point of principle. I’m not sure whether there’s a similar example in prior cases, but I’m sure Judge-elect and forever-nerd- Dapo Akande would know…
The other judges in the minority were quite vehemently opposed to this overruling of the Court’s Bosnia and Croatia judgments on this point (see Nolte, paras. 75-9; Tomka, paras. 2-28; Abraham (all of it)). Judge Abraham was particularly trenchant, concluding his opinion (likely his last on the bench) by saying that, on this point, no one could fault the Court for being excessively prudent (para. 44 – ouch).
Personally, I have no problem with the Court’s departure from its pronouncements in Bosnia and Croatia as such; there are good things to be said for both views. But the dissenters are I think right beyond any doubt when they criticize the majority for not explaining, carefully and rigorously, why the departure from Bosnia and Croatia was necessary and justified. The GvM judgment does not even acknowledge that a departure is being made, let alone explain why it is warranted. Nor does it attempt to square this departure with the Court’s own pronouncement in Croatia (2008 preliminary objections judgment), para. 53, that it ‘will not depart from its settled jurisprudence unless it finds very particular reasons to do so’, reasons that in para. 54 it also describes as ‘compelling’ (irony to be appreciated, even if the dictum was repeatedly reiterated since).
In other words, change can be good – but it needs to be acknowledged, explained and justified. That is here done only through separate opinions (on the justification side, see also Brant, paras. 11-15; Hmoud, paras. 29-33), presumably because there was no agreement in the majority on how exactly to justify this. That silence is unfortunate.
In the next (and final) instalment, I will be looking at the main question in the judgment – the proof of genocidal intent, and how the Court split on it, on change more subtle than explicit.
https://www.ejiltalk.org/did-the-court-change-its-approach-to-defining-or-proving-genocidal-intent-in-gambia-v-myanmar-and-did-it-get-it-right-part-ii/ In the first part of this post,
November 21, 2026
Did the Court Change its Approach to Defining or Proving Genocidal Intent in Gambia v. Myanmar and Did it Get it Right? Part II
Written by Marko MilanovicI looked at explicit efforts within the Court in GvM to change its approach when compared to its Bosnia and Croatia cases. In this part, I will look at the innovations that the Court made that do not depart from prior case law, but rather develop it as necessary. I will also examine how the Court reiterated prior views on standards and methods of proof, and how it then proceeded to apply them to the facts. And I will try to give an answer to the questions I posed – did the Court change its approach to defining and proving genocide, and if so, did it get it right?
Reaffirming prior jurisprudence
Much of the Court’s discussion of the definition of genocide (paras. 284 et seq.) simply restates orthodox positions previously taken by the ICJ and international criminal tribunals. So does most of the Court’s discussion of matters evidence. Thus, despite Judge Charlesworth’s invitation discussed in my previous post, the Court affirmed that proving genocide requires ‘fully conclusive’ evidence, and that the level of certainty needed depends on the seriousness of the allegation (para. 80). Contra Judge Charlesworth, my sense is that this standard of proof would under the Court’s existing jurisprudence apply equally to findings of crimes against humanity and war crimes; it is not confined to genocide, and the Court has never said that it is confined to genocide. In Bosnia and Croatia, the Court avoided using the ‘beyond a reasonable doubt’ (BARD) standard because of its association with a criminal trial, but for all practical purposes the two standards were indistinguishable.
At paras. 97 et seq. of the judgment, the Court deals with third-party fact-finders, i.e. the IIFFM and the IIMM. Regarding the IIFFM, the Court noted that it applied a lower, ‘reasonable grounds to believe’ standard of proof, which warranted some caution in relying on its findings, but also held that it found its methodology to be rigorous and on the whole even more credible than similar sources it had relied upon in prior cases (paras. 109-122). As for the IIMM, which collected evidence subject to a BARD standard, the Court decided to give its reports considerable weight (para. 119). None of this is terribly novel when compared to prior cases (and not just the genocide ones).
A welcome clarification on the relationship between IHL and genocide
There are a number of points in the judgment that provide welcome clarification on matters that were previously left open or otherwise caused confusion. I will not examine them exhaustively, but will only look at some key examples. The first is the question of the relationship between genocide and IHL, which the Court examines at paras. 296-305 of the judgment.
That question has become of critical importance especially in light of Gaza; it was practically of much less import in GvM, simply because the vast majority of killings and other violent acts at issue were done against persons already in the power of Myanmar’s military, i.e. not as part of attacks in the course of conduct of hostilities that could even plausibly be justified under the targeting rules of IHL. This is unlike the situation in Gaza, where most deaths were caused precisely by the hostilities. On this question, see further the excellent recent symposium in the JICJ edited by Paola Gaeta and Giulia Pinzauti.
In GvM, the Court clarifies its position in Croatia by reaffirming that the actus reus of killing under Article II(a) of the Convention does not require that the killing be unlawful under IHL. That is, IHL is a separate body of law and compliance with it is not relevant for the material element of genocide as an offence. (The position is likely different, in my view, for the crime against humanity of murder – for an extensive discussion of this question, see this fantastic new piece by Miles Jackson).
Rather, as the Court rightly says in GvM, (non-)compliance with IHL may be relevant for establishing either the general intent for each actus reus of genocide or the specific intent element of the offence (para. 305). This must be right. Thus, for example, if in an armed conflict a commander demonstrates that he complied with IHL rules on distinction, proportionality and precautions in attack, i.e. essentially did everything he could have done to minimize harm to civilians, this would normally preclude the finding that the commander ordering these attacks acted with genocidal intent. The two mental states simply negate each other. By contrast, one could imagine a situation in which lower-level perpetrators do IHL-compliant acts and do not have genocidal intent, but are being directed in such acts by persons who do have such intent (see Kevin Heller’s discussion here).
Pushing at the edges of the definition of genocide
As I noted above, much of the Court’s account of the definition of genocide is orthodox. In particular, the Court reaffirms the earlier position that intended destruction of the group must be physical or biological (para. 289). But the Court qualifies that position somewhat in the same paragraph by saying that:
Such destruction of the group, however, is not confined to the death or biological destruction of its individual members, or even their physical harm. This is demonstrated by the exhaustive list of material acts set forth in Article II (a) to (e) of the Convention, which are not cumulative. Acts causing serious bodily or mental harm to members of the group under Article II (b) do not, per se, involve their death or biological destruction, and serious mental harm may not involve physical harm at all. The injuries set forth in subparagraph (b) can, however, impact the ability of the group, as such, to maintain its existence as a group.
The Court then reinforces that point by examining the actus reus of the forcible transfer of children from the group, which does not entail any deaths of physical harm (paras. 290-1). The Court also notes at para. 292 that ‘[t]he absence of the term “physical” in the chapeau of Article II, however, further suggests that, for the purposes of genocidal intent, the intent “to destroy” the group conveys a broader meaning than the death of its members.’ Thus, for the Court (para. 293):
Article II of the Convention recognizes that destruction of the group “as such” may result from any of the five material acts set forth therein, whether occurring individually or in combination, when directed against at least a substantial part of the group. Thus, while the final text of the Convention excludes specific acts relating to “the destruction of historical, cultural and religious heritage” from the list of acts that must be established in order for genocide to occur (2007 Bosnia Judgment, I.C.J. Reports 2007 (I), pp. 185-186, para. 344), measures such as serious bodily or mental harm or forcible transfer of children, which can severely impact the existence of the protected group and its ability to function as a group, were considered sufficiently grave to constitute material acts under Article II of the Convention, without involving the death or biological destruction of individual members. The Court accordingly considers that the destruction of the “group, as such”, in whole or in part, as set out in the chapeau of Article II, refers to forms of destruction of the group in whole or in part that may result from the acts enumerated in Article II (a) to (e).
Is this right as a matter of principle? Yes, to the extent that Article II of the Convention expressly accepts the possibility that genocide may be committed that by acts do not cause the deaths of the group’s members, as with the infliction of mental harm or the forcible transfer of children. And there is such an enormous literature on this definition, and multiple plausible approaches to it, that I can’t say that what the Court held here is unreasonable or a real departure from prior case law. That said, note how loose some of the language is, in particular the reference to acts that ‘severely impact the existence of the protected group and its ability to function as a group,’ while simultaneously affirming that the destruction has to be physical. The immediate question this raises is how genocide can be reliably distinguished from the forcible displacement of a group from an area, coupled with many acts of murder and persecution, i.e. ethnic cleansing.
A further bit of interpretative loosening happens when the Court discusses the general intent for each of the actus rei of genocide later in the judgment. It first does so with respect to the killing of members of the group under Article II(a), at para. 315:
The Court recalls that the words “killing” and “meurtre”, appearing in the English and French versions of Article II (a) of the Convention respectively, have the same meaning, referring to the act of intentionally killing members of the group (2015 Croatia Judgment, I.C.J. Reports 2015 (I), pp. 68-69, para. 156, citing the 2007 Bosnia Judgment, I.C.J. Reports 2007 (I), p. 121, para. 186, and Blagojević and Jokić, IT‑02‑60‑T, Trial Chamber I, Section A, Judgement of 17 January 2005, para. 642). In this regard, it observes that the expression “intentional killing” has been given the same meaning as “murder” within the definition of crimes against humanity in the jurisprudence of international criminal tribunals. In the Court’s view, a killing is intentional within the meaning of Article II (a) of the Convention where the perpetrator meant to cause the death of the victim or acted with knowledge that death was a likely consequence of the act or omission. This does not include negligence or gross negligence (Karadžić, IT-95-5/18-T, Trial Chamber, Judgement of 24 March 2016, paras. 448 and 542; Blagojević and Jokić, IT‑02‑60‑T, Trial Chamber I, Section A, Judgement of 17 January 2005, paras. 572 and 642; Krstić, IT-98-33-T, Trial Chamber, Judgement of 2 August 2001, para. 543; Extraordinary Chambers in the Courts of Cambodia, Case 002/02, Trial Chamber, Judgement of 16 November 2018, paras. 635-651 and 796; see paragraph 287 above). (emphasis added)
Look carefully at the emphasized bit above – the Court is here saying that dolus eventualis (defined in terms of subjective foresight of death to the level of likelihood, rather than mere serious risk or possibility), suffices for the general intent for this actus reus (the specific intent would also need to be proven). Note also the authorities that the Court cites for this proposition, all of which are ICTY trial judgments (Karadžić appears to be the direct source of the language the Court used); there is also no ICTR judgment, and no appeals judgment cited, all of which use a probabilistic foresight formulation without much discussion in the context of murder. Plus there is an ECCC trial judgment that does have a more extensive discussion of dolus eventualis.
There is, as far as I am able to establish, very little appellate authority from the ad hoc tribunals on dolus eventualis in the context of the killing actus reus of genocide. Before the ICC, dolus eventualis would not suffice, because under Article 30(2)(b) of the Rome Statute a person has intent when ‘[i]n relation to a consequence, that person means to cause that consequence or is aware that it will occur in the ordinary course of events.’ That italicized form of intent is what is in different national systems regarded as dolus directus in the second degree, or dolus indirectus, indirect or oblique intent. The terminology varies, but generally this form of intent requires foresight to the level of virtual certainty, rather than possibility or even likelihood, i.e. it is narrower than dolus eventualis.
The practical consequence of the ICJ’s approach is that in situations involving alleged genocide committed in the course of the conduct of hostilities (think Gaza) it will be easier to establish the killing actus reus and its general intent element. A soldier who incidentally kills civilians while attacking a military objective, and who subjectively foresaw to a degree of likelihood that civilians would die as collateral damage (which often is the case) would be committing the killing actus reus with the general intent required at the dolus eventualis level. However, they would be liable for genocide only if the specific intent was also proven. (See also Paola Gaeta’s recent JICJ piece, at 2.B).
The Court then goes a step further into what I think is completely uncharted territory by holding that dolus eventualis also applies to two other actus rei of genocide:
- causing serious bodily or mental harm (‘ 334. With respect to the general intent required for Article II (b), the Court considers, at a minimum, that the above violent acts were committed by the Myanmar security forces with knowledge that they were likely to result in serious mental harm to the Rohingya population. They thus constitute intentional acts within the meaning of Article II (b) (see paragraphs 287, 315 and 320 above).’ (emphasis added))
- imposing measures intended to prevent births within the group (‘ 369. The Court observes that widespread sexual violence including the rape of underage girls and pregnant women, gang and mass rapes, and rapes in front of family members have been found to impact both the physical capacity of women to reproduce and the psychological procreative capacity of both sexes, and to constitute measures to prevent the reproduction of members of the group for purposes of the actus reus of genocide. With respect to intent, the Court considers that the likely effects of the widespread sexual violence against the Rohingya population would have been understood by the Myanmar security forces who perpetrated them, particularly in light of the cultural context, and that the acts taken must have been taken with the knowledge that these consequences were likely to occur (see paragraphs 287 and 315 above).’ (emphasis added))
The Court does not, however, do this manoeuvre regarding the Article II(c) actus reus of deliberately inflicting on the group conditions of life calculated to bring about its physical destruction in whole or in part, which it does not find was met on the facts regardless of general intent (paras. 344-356).
So, did the Court somehow redefine genocide? No, not really. But is there some interpretative loosening here? I think that is reasonably clear. The looser language used to describe the physical destruction of the group as part of the specific intent, and the use of dolus eventualis for three of the four actus rei of genocide, do make genocide somewhat easier to establish than was previously the case, or was previously assumed. For example, the infliction of serious bodily or mental harm in hostilities, subjectively foreseen to the level of likelihood – something that’s inevitable in most conflicts – would suffice for the Article II(b) actus reus.
I really have no opinion on whether this is a good idea or not. The specific intent element of genocide is in my view so arbitrary that debates about it are not always fruitful, if I can put it that way. If one’s starting point is that as clear a line needs to be maintained between genocide and ethnic cleansing, as was the view of the minority, then the assessment of the joint dissent (at para. 8), ‘certain formulations [in the judgment] give rise to concern’ – is correct, but no more than that. The Court’s actual approach to analysing the evidence was far more impactful.
I am somewhat surprised by the introduction of dolus eventualis as the general intent requirement for three of the four actus rei, and a bit more surprised by the fact that (unless I’ve missed this in haste) none of the separate opinions discuss it, other than a brief mention in Judge Tladi’s declaration, at paras. 23-4.
Recall that in his separate opinion Judge Tladi argued for the use of dolus eventualis for the specific intent element of genocide. Thus, to use a classical example, if I were to drop a nuclear bomb on Belgrade, I would in Judge Tladi’s view presumably commit genocide against the Serbs (the part of the Serb population who live in Belgrade, a couple of million people), regardless of the reason why I did so, because the death of tens or hundreds of thousands of Serbs would be something that I subjectively foresaw to the degree of likelihood, and accepted that and dropped the bomb nonetheless. Genocide would not require me specifically wanting to destroy Serbs as a group. By contrast, the use of dolus eventualis as the general intent element for the actus rei simply means that the underlying conduct elements are easier to satisfy, especially in the conduct of hostilities context, but that has never the most difficult thing about proving genocide.
A much more flexible approach to the evidence
The main determinant of the outcome of the case was the Court’s more flexible approach to evaluating the circumstantial evidence probative of genocidal intent. On the bare facts, when it came to proving atrocity crimes in specific places, the Court relied extensively on the findings and evidence of the IIFFM and the IIMM, in a fashion that is broadly similar to how it relied on the ICTY in Bosnia and Croatia, even though the overall number of incidents was much smaller. The real issue was evidence of genocidal intent, and there was no direct evidence of such here.
The Court’s basic posture was that it would not infer genocidal intent from a pattern of conduct, but rather from the totality of the evidence (paras. 387-8, 396). All of this evidence is circumstantial. Frankly I think this is hard to disagree with, even if the joint dissent does so to some extent. The issue is not whether the totality of the evidence/all of the evidence taken together is the right reference point, but with the factors that the Court actually choose to use (para. 397):
Accordingly, the Court will base its assessment of the existence of genocidal intent in the present case on the following factors: (a) the general context, including the perpetration of other culpable acts systematically directed against the Rohingya; (b) the selective targeting of the Rohingya; (c) the perpetration of acts that violate the foundations of the Rohingya group; and (d) the systematic nature and the scale of Myanmar’s actions.
The problem with these factors is simply that they are not unique to genocidal intent. I won’t reproduce here the Court’s analysis of these factors; suffice it to say that while many of the circumstances it relies on permit the inference of genocidal intent, they also permit other inferences (e.g. to forcibly displace the Rohingya). And this really was the key question in the case – whether a finding of genocide against the Rohingya was the only reasonable inference (a point I will come to below), or whether the atrocities committed against them had a reasonable alternative explanation. After reading the Court’s analysis, I found it hard to see how the exact same approach would not have led to a different outcome in Bosnia in particular – not necessarily in the Court finding that the totality of the conflict in Bosnia was genocidal (as was the applicant’s argument), but in finding that genocide was also committed in other municipalities, such as Prijedor.
That said, some aspects of the Court’s GvM analysis are both undoubtedly correct and will have a direct impact on future cases:
- 429: ‘The Court observes that, while a high relative number of deaths may well be an important factor from which to infer genocidal intent, a comparatively lower percentage of deaths does not, by itself, necessarily preclude a finding of genocidal intent.’ This must be true, but it is also true that scale of killing in GvM was also not such to obviously assist the finding of genocide.
- 430: ‘a reasonable inference of genocidal intent does not depend on evidence that the perpetrators made every effort to maximize the level of harm that they caused to the protected group.’ Again, this must be true – there are very good reasons (including avoiding condemnation, sanctions etc.) why a perpetrator may refrain from maximizing harm. (Think, again, here of arguments on whether Israel committed genocide in Gaza.)
The Court thus concluded that:
- The Court’s examination of the above factors establishes that the context surrounding the 2017 clearance operations, including decades of discriminatory legislation and a military-sponsored hate speech campaign against the Rohingya, created an atmosphere that was conducive to the violence perpetrated during those clearance operations by Myanmar’s security forces and formed the basis of Myanmar’s intention to destroy part of the Rohingya group. The manner in which the forcible displacement of the Rohingya was conducted, together with the selective targeting of the Rohingya population for violence and the burning of their homes, which sent a clear message that they had nothing to return to, supports the inference of an intention to destroy the group. The evidence of brutal acts of sexual violence and the targeting of children for violence in Rohingya communities also supports an inference that there was a desire to destroy the foundations of the group. The systematic nature of these actions further indicates a reasonable inference that Myanmar acted with genocidal intent.
This is all sensible. The problem with this finding is that it is not ‘fully conclusive’ in the way in which that standard of proof was applied previously. Literally all of the factors the Court mentions here (other perhaps than the decades of discriminatory legislation) apply to dozens of situations all over the world that are commonly regarded as ethnic cleansing/crimes against humanity, including during the war in Bosnia.
The Court therefore found that genocide intent was a reasonable inference from the evidence (para. 437). It then proceeded to establish whether it was the only reasonable inference. Unlike the joint dissent, I see no problem in this two-step inferential approach. The whole idea of an ‘only reasonable inference’ presupposes that there are situations in which there are multiple reasonable inferences available. Some may, of course, be more persuasive, while unreasonable inferences are clearly not to be considered. Thus, the question for the Court was whether genocidal intent was the ‘only’ reasonable inference available.
A welcome clarification on proving genocide and the ‘only reasonable inference’ test
I have extensively discussed this ORI test in a previous post, the substance of which I won’t repeat here. Suffice it to say that the test (1) is not at all specific to genocide, and has been applied in dozens of cases that have nothing to do with genocide and (2) that the ‘only reasonable inference’ of intent does not mean there has to be only one intent, one reason for acting. The ORI test has also been the subject of much dispute between the parties to the case and the intervening states. The degree of confusion on these points has been gobsmacking. The Court has now provided some much needed clarification.
The Court thus holds, entirely correctly, that:
- 442: ‘The “only reasonable inference” standard does not derive from the Genocide Convention itself but is a general test under international criminal law, applying whenever inferences are drawn from circumstantial evidence.’ The only thing missing here is that the reason the ORI test is applied is because facts (including intent) need to be proven beyond a reasonable doubt. That is, if there are alternative reasonable inferences, there is a reasonable doubt remaining and BARD is not satisfied. And ‘fully conclusive evidence’ is in this context analogous to BARD.
- 442: ‘both the Court and international criminal tribunals have recognized that genocidal intent may exist alongside other intentions, provided that such intentions are not incompatible with the existence of genocidal intent.’ Again, quite clearly correct – ORI is not about identifying the only intent. A person (or a state through a person) can act with multiple intents. The question is what intents are compatible with genocidal intent.
- 450 – the applicant bears the burden of proof; the respondent does not bear the burden of showing there are alternative reasonable inferences. Having said so, the Court then looks at three possible alternative reasonable inferences – precisely those that Myanmar had identified in its argument, and not any others.
The first is that Myanmar was conducting a counter-insurgency operation against ARSA, a Rohingya armed group. The Court finds that, while there was a correlation between the clearance operations and ARSA’s activities, Myanmar made no effort to spare civilians. Much of the violence was directed specifically at civilians outside combat. Thus, at para. 453, ‘the Court concludes that the intention of conducting genuine counter-insurgency operations does not constitute an alternative reasonable inference’ (emphasis mine). Note how the word ‘genuine’ here is used to qualify the description of the alternative intent.
This is I think correct, and is relevant directly for weighing which further intents are compatible with genocidal intent, and which are not. The broader proposition, which the Court does not consider here expressly, is that an intent to win the war, including by deliberately harming civilians, is perfectly compatible with genocidal intent. That is, in defining the enemy, and defining what it means to defeat or suppress that enemy, a party to conflict can (and often does in practice) use violence against civilians as a deliberate strategy. That violence can include genocide; the genocide might not be a side-effect of fighting and trying to win the war, but one of its main tools and intended outcomes. Note the obvious relevance of this point to the question whether Israel committed genocide in Gaza – ‘we were intending to completely defeat Hamas’ is not an intent that is per se incompatible with genocidal intent. However, genuine, sustained efforts to minimize harm to civilians are incompatible with such intent.
The third alternative intent (I will come back to the second in a moment) is that of collective punishment. This intent was understood by the Court so ‘that Myanmar is suggesting that its security forces might have intended to punish the Rohingya civilian population in retaliation for the activities of ARSA’ (para. 463). The Court then reasons that:
- … [G]enocidal intent may co-exist with other intentions, as long as such intentions are not incompatible with the existence of genocidal intent (see paragraph 443 above). In the Court’s view, genocidal intent is likely to be accompanied by an intent to punish, particularly in the context of collective punishment. Genocidal and punitive intentions may often be entwined in the sense that the destruction of a protected group is an extreme form of punishment, and the two intentions may not always be capable of meaningful separation for the purpose of drawing alternative inferences.
Again, I think that’s entirely right. In the Srebrenica genocide, for example, the genocidal intent was undoubtedly accompanied by the intent to take revenge on and punish the population of Srebrenica for prior attacks by Bosniak armed forces against Serb civilians and villages in the enclave’s vicinity. There is no incompatibility between these intents. Similarly, to go back to the Gaza example, an intent to collectively punish the population of Gaza for supporting Hamas, or for failing to stop Hamas in acts against Israel and Israeli civilians, would be entirely compatible with genocidal intent.
Where the Court’s analysis falters, however, is in its engagement with Myanmar’s second alternative – ethnic cleansing/forcible displacement (paras. 454-462). The Court first says that an intent to displace the population can, in some situations, be compatible with an intent to destroy the group. Srebrenica is one such example, where women, the very young and the elderly were allowed to leave or were evacuated, whereas the men and boys were killed (note, however, that is precisely the reason why Srebrenica was not an obvious case of genocide, unlike say Rwanda). The Court at some length attempts to distinguish GvM from its finding in Bosnia that the ethnic cleansing campaigns conducted by the Bosnian Serbs in numerous areas of the country were not genocide. It then concludes that
- The Court thus considers that an intent to forcibly displace the population does not explain the totality of the evidence before the Court. A number of aspects of the clearance operations, including the brutal manner in which they were conducted, attacks by the security forces on the Rohingya as they fled to Bangladesh and the widespread attacks on young children, are not consistent with an intent to forcibly displace the population. For this reason, in the Court’s view, an intent to forcibly displace the population does not constitute an alternative reasonable inference.
I do not find this at all persuasive. All of these factors existed in numerous places in Bosnia as well. That Myanmar did not organize ‘evacuation’ buses, the way that the Bosnian Serbs often did, does not mean that their overall plan was not to push the Rohingya population into Bangladesh, even if not every single act by every single Tatmadaw commander on the ground was consistent with that overall plan. The Court itself acknowledges (para. 262) that ‘[o]ver 700,000 Rohingya, representing over 85 per cent of the Rohingya population who remained in northern Rakhine State after the events of 2016, fled to Bangladesh as a result of the 2017 clearance operations.’ Brutalizing civilians, including by killing some of them, by burning their villages or by means of sexual violence, is precisely how they are compelled to flee – again exactly what happened in many parts of Bosnia or in Kosovo in 1999.
To be clear, I am not saying here that one cannot reasonably hold the view, on the basis of the evidence presented, that Myanmar (i.e. its officials and soldiers) acted with genocidal intent. What happened is consistent with genocidal intent. What I am saying is that this is not the only reasonable inference available. That is, one can equally reasonably hold the view that what happened in 2017 is explainable by an overall policy to displace the group, rather than to destroy it physically or biologically. Maybe one of these views is a better representation of the evidence, but on the basis of the IIFFM/IIMM reports and the evidence presented in the judgment itself, both appear to be reasonable. The fact that we all knew about all of this evidence and were still not sure whether Court would find genocide tells you that the alternative inference is reasonable!
Genocide or bust
So, to answer my two questions. Did the Court change the definition of genocide? Not really, even if it tinkered with it at the margins. Did the Court change its approach to proving genocide? Yes. Some of that change was simply normal development and clarification, so not really ‘change’ as I defined it. But some of it is a sub silentio, de facto relaxation of the ‘fully convincing evidence’ standard of proof, while the Court nominally remained committed to it. The joint dissent’s criticism of the majority on this point is fair.
That said, this is not as unusual or radical as it may sound. A standard of proof is primarily a self-disciplining tool for a trier of fact, which tries to streamline their reasoning process. Even when agreeing on a high standard of proof, judges may evaluate the evidence flexibly – the one example I would commend to a reader would be how the ICTY/MICT drew inferences about Radovan Karadzic’s genocidal intent regarding Srebrenica, and did so on a standard of proof beyond a reasonable doubt. And sometimes judges talk the language of a standard such as BARD but in practice are nowhere close to applying it – the European Court of Human Rights is the best example here. Finally, and obviously, whatever one says about the applicable standard of proof it cannot, in practice, be applied identically after a trial that lasted 15 trial days and a criminal one that lasted 500. These are just wholly different genres.
This is the descriptive question. I don’t think it is a hard one, in the sense that a holistic reading of the judgment based on the totality of the evidence (!) leads one to a reasonable inference that the Court’s approach has become more flexible. The hard question is the normative one – was this the right thing to do?
Here there will be much disagreement, resting, as I explained in my earlier post, on various priors. I will briefly set them out here, without further elaboration: (1) in cases brought under the Convention the ICJ’s jurisdiction is limited to genocide, which means that it cannot pronounce on crimes against humanity and war crimes; (2) that creates an incentive to litigate a very complex situation or an entire conflict solely through the prism of genocide; (3) the public stigma attached to genocide has far overshadowed that attached to crimes against humanity and war crimes; (4) this in turn leads to competitive victimhood, to a group’s active desire to be seen as a victim of genocide, with that label and only that label being sufficient to encapsulate harms caused to it; (5) this then leads to a widespread understanding of genocide as justice and success, and no genocide as injustice and failure; (6) the Court’s judges thus face a combination of internal and external pressure to accommodate this dynamic. Some want to preserve genocide on a pedestal of condemnation and reserve that label only for the most obvious of cases, others think this is far too restrictive and makes the Genocide Convention a dead letter.
Speaking for myself, I have long been a genocide-allergy sufferer. I deplore its public elevation over other international crimes. In the former Yugoslavia, the competitive victimhood around that word, which started even before Yugoslavia fell apart and unfortunately persists to this day, makes me wonder whether we would have been better off had Lemkin never invented the term. I still remember the day on which the Court delivered its Bosnia judgment (I clerked there at the time), when a Bosniak activist tore the judgment apart at the gates of the Peace Palace because the Court found that ‘only’ Srebrenica, and not the totality of the war, was genocide. The concept of genocide as defined in the Convention, and the case law that interprets it, are so pervaded by moral arbitrariness and so poisonous in the narrative battles they provoke, that I just want to wash my hands off it. So I will leave the reader with this 2024 New York Times piece by Linda Kinstler on the bitter fight over genocide. I would only add a parting comment that, despite my criticism of the Court’s ultimate conclusion in GvM, the Court as a whole (including the dissenting judges) did some admirable work here, and much of the Court’s reasoning will be relevant for decades to come.
No comments:
Post a Comment