This week, a UN commission of inquiry found that Israel is committing genocide in Gaza. NGOs and various other bodies have of course made that finding before. Proving genocide, in Gaza or elsewhere, invariably turns on proving the specific intent of the perpetrators. In this post, I do not want to get into the Gaza genocide debate or discuss any particular item of evidence. Rather, I will examine one particular myth, if I can call it that: that there is some kind of special, legally mandated evidentiary rule of proving genocidal intent, so that such intent must be the ‘only reasonable inference’ (ORI) available.

There is a lot of confused commentary on this issue. There is also some confusion in the CoI report on Gaza in this regard, as I will explain below. But the correct legal position is actually quite simple: 

  1. ORI is an evidentiary rule that is not in any way specific to genocide. It applies across the board in criminal trials when a certain fact, which is necessary to prove an element of the crime, is inferred from circumstantial evidence. It is not limited to facts probative of intent. 
  2. ORI is a result of the high standard of proof required for conviction in an international criminal trial – beyond a reasonable doubt (BARD). In other words, there is a direct link between ORI and the degree of certainty that a fact-finder must have, which is context-dependent.
  3. ORI was adopted by the ICJ in the Bosnian and Croatian genocide cases because the Court there adopted a high standard of proof, which is more or less equivalent to BARD (to my mind, rightly so). 
  4. This high standard of proof, and the consequent reliance on ORI, is again not specific to genocide – it would equally apply to war crimes or crimes against humanity – but it does not apply outside the courtroom. Nor does it apply to other contexts within the courtroom, e.g. the issuance of arrest warrants or the confirmation of charges, that precede trial and conviction.
  5. Other fact-finders are not bound to apply a high standard of proof or ORI, for genocide or other international crimes alike. This includes formal fact-finding bodies, like a UN CoI, but also governments, NGOs, academics, or ordinary people. One can, for example, rationally choose to believe that Israel is committing genocide or crimes against humanity in Gaza without requiring proof beyond a reasonable doubt and without applying ORI.

Let me develop some of these points in more detail.

ORI, BARD and ICTY/R/MICT case law

The notion of an ‘only reasonable inference’ comes to international law from the jurisprudence of the first modern ad hoc criminal tribunals. That jurisprudence is crystal clear that ORI is a function of the very demanding standard of proof for conviction – beyond a reasonable doubt – which entails that the judge is convinced to a very high level of certainty that a particular fact is true. The judges of the ICTY/R and the MICT applied this notion in all sorts of cases, most of which had nothing to do with genocide. ORI was sometimes applied to establish mens rea, sometimes facts relevant for the actus reus. This is not a rule about proving intent, specific or not, but a rule on how one should evaluate circumstantial evidence.

So, for instance, in the Celebici case the judges had to establish whether one of the defendants, Delic, participated in the beating of a victim. There was no direct evidence of him doing so. The issue was whether an inference could be made about his participation from circumstantial, indirect evidence. This is how the Appeals Chamber put it (para 458):

A circumstantial case consists of evidence of a number of different circumstances which, taken in combination, point to the guilt of the accused person because they would usually exist in combination only because the accused did what is alleged against him – here that he participated in the second beating of Gotovac. Such a conclusion must be established beyond reasonable doubt. It is not sufficient that it is a reasonable conclusion available from that evidence. It must be the only reasonable conclusion available. If there is another conclusion which is also reasonably open from that evidence, and which is consistent with the innocence of the accused, he must be acquitted.

Note the express link drawn by the judges between the BARD standard of proof and the necessity that the inference be the only reasonable one – if there are other inferences (conclusions) available, and these are reasonable, then the conduct or guilt of the accused have not been proven beyond a reasonable doubt. Note also the underlying idea, which is that one can have different degrees of confidence about facts said to have existed in the past, i.e. that establishing the truth can be a matter of degree, but that the degree of certainty required will vary depending on the context and the consequences.

Again, the tribunals applied this approach in all sorts of cases, most of which were not about genocide – readers can find a useful summary of this jurisprudence here, and it is likely not a comprehensive one.  

ORI is not specific to genocide

I hope that this point is now obvious. ORI applies if BARD applies, and it is relevant for proving any fact necessary to establish an element of any crime from circumstantial evidence. The death of a particular person, for example, whose body has never been found, could be so established – say if the person was last seen in the custody of some soldiers, was seen to be headed in the direction of a mass execution site where many other similar victims were killed, the bodies of those killed were disposed of, and the person never reappeared after the event. From those facts one could infer the death of this person – and so forth – when the direct evidence of their death, such as finding their remains or the testimony of someone who witnessed the killing is unavailable.

Similarly, if a commander was charged with the war crime of intentionally directing attacks against civilians, which for conviction requires proof BARD that the commander knew that he was attacking civilians and not combatants/military objectives, the commander’s knowledge and intent would generally need to be inferred from circumstantial evidence, such as patterns of conduct. If some other reasonable explanation was available – for instance that the commander honestly believed that he was pursuing military objectives and simply made a mistake – he would have to be acquitted. There’s nothing genocide-specific here. It is simply that reasonable doubt can continue to exist if other reasonable conclusions (inferences) are possible.

The ICJ’s adoption of ORI in the two genocide cases

As noted above, the ICJ adopted the ORI approach in evaluating evidence of genocidal intent in the two genocide cases it has so far decided on the merits – see Bosnia v. Serbia, para 373 and Croatia v. Serbia, para 148. But this was entirely the result of the high standard of proof the Court required – fully conclusive/convincing evidence (Bosnia v. Serbia, para 209). While the Court did not frame that standard of proof as BARD, for practical purposes it was indistinguishable from BARD. Basically, the judges were not comfortable with using the BARD terminology, because it was evocative of criminal proceedings, because many of them came from civil law systems that do not employ formal standards of proof, and because they wanted some flexibility in future cases – but again for all practical purposes the standard of proof required was equivalent to BARD when it came to proving the existence of genocide.

Almost twenty years ago, I argued on the pages of EJIL that such an approach to the standard of proof was entirely appropriate (see here and here). I still believe this to be true. Why? Because it would be highly problematic to have inconsistent judicial determinations of what is an international crime – genocide – depending on whether the court making the determination was a criminal or a ‘civil’ one. For example, it would be normatively undesirable for the ICTY to establish that genocide happened ‘only’ in Srebrenica, but for the ICJ to rule (after a much less rigorous evidentiary process, and one largely derivative of what the ICTY established anyway) that genocide was committed elsewhere in Bosnia as well. There is no such thing as ‘civil’ genocide – there is only the crime of genocide, for which both individuals and states can be held responsible, the former criminally, and the latter for an internationally wrongful act of a person who acted on the state’s behalf.

But the same applies for war crimes or crimes against humanity. If there was a case before the ICJ, on Gaza or something else, where the existence of these crimes was at issue, the approach to evidence should in my view be no different than with genocide. These crimes are no less grave. The evidence of such crimes would still need to be fully convincing/conclusive. Inferences of intent or other facts would still need to be made relatively strictly. ‘Genocide is special’ is really the wrong message to send here, and not something that the ICJ ever said.

ORI simply does not apply if the standard of proof is lower

Now this is the key point: one can resort to inferences much more liberally if the standard of proof lower. Or, to put this differently, ORI is directly linked to BARD, and if BARD (or fully convincing/conclusive evidence) is not the relevant standard of proof then requiring ORI is not appropriate either. This is true before criminal tribunals, but also before other kinds of judicial and non-judicial bodies. 

So, for example, to issue an arrest warrant, an ICC Pre-Trial Chamber needs to be satisfied that there are reasonable grounds to believe that the person in question committed a crime (Art 58 of the Rome Statute). This applies to all elements of the crime. Thus, if an arrest warrant is sought for the war crime of starvation, the Prosecutor needs to convince the judges that the person in question intended to starve civilians as a method of warfare – the mens rea of the crime. That intent can be inferred from circumstantial evidence. But, because the standard of proof is ‘reasonable grounds to believe’ and not BARD, that inference can be made even if there are other reasonable inferences available. Genocide is no different. To obtain an arrest warrant for genocide, one does not need to prove that the ‘only reasonable inference’ from the facts is that the person in question had genocidal intent.

I hope this is clear. If BARD doesn’t apply, then ORI doesn’t apply either, whether it’s genocide or something else. Note that the ICC arrest warrant reasonable grounds standard is generally regarded as equivalent to ‘reasonable suspicion’, a standard commonly used domestically and also found in human rights law to justify deprivations of liberty. It is quite deliberately a low standard – well below likelihood, let alone certainty. It would be simply absurd to apply ORI in that context. The same goes, for instance, for the confirmation of charges stage before the ICC, where the standard of proof is substantial grounds to believe (Art 61 RS) – higher than ‘reasonable grounds’, but well below BARD.

The standard of proof in the Gaza CoI report

This brings me to the recent report finding that Israel is committing genocide. A UN fact-finding mission is not a court, let alone a criminal court. It would be wholly unreasonable to expect of such a mission to establish facts to the BARD degree of certainty – something that often takes years of trial proceedings in the criminal context. That’s why UN fact-finding missions have never adopted BARD. That’s also why the OHCHR guidance on such missions expressly says that these bodies are not criminal courts, even if they need to articulate the standard of proof they are applying (see here, at 62-3).

One can, however, often find quite a bit of conceptual confusion on these matters of evidence in various fact-finding reports – and this week’s report is no different. First, at para 7 of the report, the Commission says that it will apply ‘the same methodology and standard of proof previously adopted for its investigations, that is, ‘reasonable grounds to conclude’, in assessing the factual findings and conclusions.’ But then, in para 156 of the report, the Commission quotes the ICJ’s fully convincing/conclusive standard of proof, and in para 220 says that

On the basis of fully conclusive evidence, the Commission finds that statements made by Israeli authorities are direct evidence of genocidal intent. Additionally, on the basis of circumstantial evidence, the Commission finds that genocidal intent was the only reasonable inference that could be drawn based on the pattern of conduct of the Israeli authorities.

This is then repeated at para 252 of the report, where the Commission ‘concludes on reasonable grounds that the Israeli authorities and Israeli security forces have committed and are continuing to commit the following actus reus of genocide’, while saying in para 254 ‘that genocidal intent was the only reasonable inference that could be drawn from the totality of the evidence.’

This is just internally contradictory. If the standard of proof is ‘reasonable grounds’, then it’s not ‘fully conclusive/convincing’, and one can definitely draw inferences even if other inferences are reasonably available. If the Commission meant to say that for proving intent only (and not for the actus reus of genocide) it was applying a higher standard of proof, then it should have explained itself more clearly. (As I read the report, that’s not what it actually did, nor do I see any normative justification for distinguishing between different elements in terms of the quantum or quality of evidence required.)

The Commission was, in my view, not at all bound to establish that Israel is committing genocide to the same standard of proof that the ICJ requires for a final judicial determination. It was perfectly entitled to use a lower standard of proof (again, from my reading of the report, that’s what it did). It was absolutely not bound to use ORI – nor are other fact-finders, such as human rights NGOs, so bound, whether they are making determinations of genocide or of other international crimes. The Commission likely used this language to preclude criticism that it was ignoring ICJ jurisprudence and the like – but as I explained above that jurisprudence is simply not what governs these other fact-finding processes. What I find especially problematic with some of these reports is the implication that genocide requires more proof and is more grave than ‘mere’ war crimes or crimes against humanity. It doesn’t, and it isn’t. 

Conclusion

The same goes for determinations by governments, other institutions, professional lawyers or other individuals. If I have to make up my own mind on whether Israel is committing genocide, crimes against humanity or war crimes in Gaza, I absolutely do NOT have to employ BARD, ORI, or any formal standard of proof whatsoever – so long as I rationally engage with problems such as uncertainty in proving intent. If, say, the Spanish government makes a determination that Israel is committing war crimes, extermination or even genocide, it is not bound by any formal standard of proof in doing so for whatever purpose it needs to do so, such as suspending armed exports, implementing sanctions, or even making a mere public accusation.

Similarly, if an international lawyer is asked whether, in their opinion, Israel is committing genocide or starvation or whatever, they are not somehow professionally bound to engage in mental gymnastics about ‘only reasonable inference’ and such. It suffices entirely from them to say that, on their best understanding of the available evidence, Israeli officials have committed genocide – or not. (Note that this is a different question from predicting whether, after a full trial, the ICJ or some other court will, in applying ORI, find that genocide or whatever other crime was committed. It is perfectly rational to believe that Israel is committing genocide or some other crime but that definitively proving so in court might be difficult.) People, institutions or governments often have to make decisions under a cloud of uncertainty while events are still ongoing. Expecting fully convincing evidence or proof beyond a reasonable doubt of the kind one gets after months or years of an adversarial trial is nothing less than a category error.