In this three-part series I seek to draw attention to legally-unprecedented and epistemologically-unsound evidentiary standards emerging at the ICC, particularly in the Gbagbo case. The mainstream reaction to the Gbagbo case has been to accept the narrative that the problem lies entirely with evidence. However, when the majority derides the “questionable quality of much of the evidence” (§1608), it speaks from a lens of Cartesian standards. If one reads the judgment instead through the lens of more typical legal standards, the evidence is harrowing. Thousands of diverse items of evidence – eye-witnesses, videos, insiders, experts, and forensic and documentary evidence – attest to hundreds of instances of killing, wounding, raping, torturing and burning of civilians by police and other pro-Gbagbo forces.
At Nuremberg, Robert Jackson warned that giving the defendants an unfair trial would be a poisoned chalice for the tribunal itself. My concern is that opposite extreme is also a poisoned chalice. An exclusive focus on the interests of the accused, to the exclusion of all other considerations, leading to rarified and ungrounded standards, will also collapse the system. If unchecked, these standards can only lead to repeated collapses of investigations and prosecutions. We are at an interesting moment, because scholars are rightly warning against ‘crisis narratives’, and I myself have appealed for less alarmism. Nonetheless I think that evidentiary standards are now one of the most crucial topics for study and reform.
The previous two posts (see Part I here and Part II here) gave only a cursory outline of problematic approaches to evidence and examples thereof. I will now touch on two related points, (1) evidentiary expectations for crimes against humanity and (2) investigative criticisms that overlook the applicable legal regime, and then I will conclude.
New barriers for crimes against humanity cases
The Gbagbo majority advances novel evidentiary expectations for crimes against humanity, which are rooted in the Cartesian desire for granular detail, certainty, and clinical clarity. I will give a very short overview here of some examples, and why they are problematic.
Pattern non-recognition: The hypersceptical atomistic approach, which I discussed in the first post, is particularly ill-suited for crimes against humanity cases. I gave examples of how the Gbabgo majority looked at each incident in isolation and advanced speculations as to how each one might be a random event unconnected to any pattern. It thereby dismissed each incident one by one. However, this method will never recognize a pattern, because a single data point in isolation cannot prove a pattern, and thus each incident gets dismissed until none are left. By contrast, the normal approach in criminal law, and in all fields of inquiry including science, is to look holistically and dispassionately at all of the evidence. Even if we are not sure which particular incidents might be outliers, we can have more than enough incidents to be confident of the pattern as a whole. For example, scientists cannot show that any particular storm in isolation is caused by climate change, but nonetheless they have extremely high confidence in the overall pattern – a clear increase in storms.
Perpetrator motives: It is well-established law (Tadic, Krnojelac, Kunarac) that proof of the motive of perpetrators is not required in crimes against humanity; they may participate in an attack for a variety of motives, some of them personal. Despite this, the Gbagbo majority repeatedly required evidence that physical perpetrators were motivated by a policy and/or had discriminatory intent. The majority claimed that motive was merely a ‘factor’ and not a requirement (§1389-91). However, in their analysis, they dismissed several incidents because perpetrator motive was not adequately proven (see eg §1415, 1487, 1708, 1742). Hence it was in fact elevated to a requirement. The requirement is unsupported in law. It is also unsound criminologically, because of its implausibly tidy conception of front-line perpetrators as purely executing instructions. (The majority acknowledged possible mixed motives when it introduced its test, but in applying the test, the majority excluded instances of mixed motives – see my next point and see §1708). It will usually be impossible to prove personal motivations when individual identities of members of an attacking group are largely unknown. Despite the difficulties of proving motives of numerous perpetrators, the Prosecution actually did obtain and present some such evidence, such as perpetrators chanting slogans or referring to instructions (see Carbuccia dissent at §146 and 162), but the evidence still did not meet the majority’s standards.
Rape as opportunistic. One of the most disturbing features of the Gbagbo majority approach is that it treats rape as opportunistic, and hence requires additional proof to acknowledge it as part of the attack. In response to mass rapes of protestors, the majority held that there was “no obvious connection” to the other repression (§1217). However, the patently obvious connection is that these mass rapes were carried out pro-Gbagbo forces, against Outarra-supporting demonstrators, in a context of denigration and permissive violence directed against that very group. The failure to see this obvious connection reflects a misconception that sexual violence is inherently different and separate from other Article 7(1) crimes.
In another incident, where multiple women were raped after being identified as Ouatarra supporters, the majority was still unconvinced: the “question in those instances is whether the identification was the reason for the perpetrators to rape their victims or whether this served merely as a pretext” (§1882). That additional question is deeply misguided. Whether a perpetrator is obeying orders or taking advantage of coercive circumstances to carry out Article 7(1) inhumane acts against a targeted group, that is in law unambiguously part of the attack.
In another incident, multiple female protestors, after being identified as Outarra supporters, were raped by pro-Gbagbo forces, next to the police station (§1467). Again, the majority was still unconvinced, this time because there was “no information about whether these rapes were committed pursuant to an instruction or … agreement” with the police (§1467). However, proof of specific instructions is never required; nor is proof of a policy to rape. In the drafting of the Rome Statute, Article 7(2)(a) was changed from “policy to commit such acts” to “policy to commit such attack” specifically because of a concern by the Women’s Caucus that judges might wrongly think that there must be a rape-specific policy. Yet despite that change and despite longstanding efforts to educate international judges, these preconceptions about sexual violence being opportunistic and unconnected still unfortunately surface.
Consistent organization-wide implementation. An entirely novel and and perplexing feature of the majority approach is the idea that there should be consistent implementation by all members of an organization in order to find a policy. No legal support at all was advanced for this innovative thought. Article 7(2)(1) merely requires ‘multiple’ inhumane acts, with sufficient coordination to infer an underlying state or organizational policy, as opposed to random, isolated, unconnected individual acts. The suggested new benchmark of organization-wide implementation is far more onerous even than ‘”systematic”. It is a recurring and regrettable trend that some ICC chambers keep inventing new onerous requirements rather than simply applying the established law.
Because of this novel benchmark of consistent implementation, the fact that some victims were detained and beaten extensively, rather than killed, was taken as undermining the evidence of crimes against humanity (§ 1560-1562). In another incident, the majority found the fact that the victim was merely “beaten by armed individuals” (§1593, emphasis in original) to be significant, because being armed they could have killed him. The fact that some policed warned marchers that irregulars would harm them was thought by the majority to undermine the evidence of policy (§1610), whereas it actually it proves the policy: it shows coordination and knowledge of what was about to ensue.
This new benchmark of consistent implementation led to a particularly novel argument. The majority argued that even on a holistic assessment, there was no pattern, because one must consider the entire population of the city, and all of the interactions between the police and the target groups. In other words, “look how many people the police didn’t kill!”. The majority argument has merit in one way: a small number of isolated harms, in a large country, spread over years, would not constitute an ‘attack’. But deliberate killing or harming of 500 or more unarmed civilians in a single city in a span of a few months is more than enough to constitute an ‘attack’.
The attempted comparison with the size of the entire potential victim group wrongly conflates crimes against humanity with genocide. Furthermore, the majority is injecting requirements more demanding than ‘widespread and systematic’ into the simple, modest requirements of an ‘attack’. An attack does not require assaulting all possible victims in a target community, nor does it require all members of a group acting at all opportunities to harm people. To help guide one’s understanding, it might help to think for a moment about the common sense and ordinary meaning of ‘attack’. Under Article 7, a single attack with “multiple” victims suffices; then we proceed to the more difficult tests of “widespread or systematic”.
Judges Henderson and Tarfusser accuse the Prosecutor of “cherry picking” because she presented evidence focusing on criminal incidents (§1888). This line has also been echoed in commentaries, but I think the criticism is unsound. It is not “cherry picking” for a prosecutor to present evidence of apparently criminal incidents. On the contrary, presenting evidence of possibly criminal incidents is a prosecutor’s mandate. The ‘cherry picking’ objection is rooted in these misunderstandings – the expectation of consistent implementation and the incorrect genocide-style comparison. You don’t have to show that the police tried to harm people in all or most interactions with a target group. That is not even required for genocide or for systematic attacks.
Overlooking the legal regime: the OTP is not the police
It is entirely appropriate for judges (and others) to criticize faulty investigative practices. However, in the ICC context, criticisms often show unfamiliarity with the relevant legal regime (for example Part 9) and mutual legal assistance. One possible example is the reprimand from Judge Tarfusser: “I saw with disbelief that, instead of formally seizing original items, the Prosecutor and her investigators had simply photocopied them” (§91). This criticism has been echoed in commentaries, but I invite readers to pause to reflect critically. When you go into the governmental archives of a sovereign state, you don’t generally get to take away original documents. This is standard practice in national and international cooperation.
All too frequently, the ICC is criticized for not following familiar steps that national police would follow, while failing to take into account that the OTP is not the police. It does not have the legal powers, resources, or staffing levels of national police. The ICC is an extraordinary mechanism, and its investigative powers in the Statute are heavily modelled on a “mutual legal assistance” approach. ICC investigators have quite limited powers under the Statute, and additional powers as negotiated with territorial states. ICC officials often must arrive “late”, because of the admissibility regime, and they rely on cooperation partners who in turn might have limited resources and face chaotic situations. Criticisms of the OTP for not immediately securing crime scenes or evidence often fail to take into account the relevant legal regime and limitations. In addition, the forensic standards expected of national police when they deal with a single crime may be entirely inappropriate for crimes of massive scale.
My aim is to draw attention to well-intentioned but unprecedented evidentiary standards that need to be studied. In my view, these must be corrected if the ICC is to be a viable forum and carry out its social mandate. I have only provided a preliminary sketch here.
The Cartesian approach would significantly increase the costs and durations of each investigation and prosecution, because of the need to collect and submit direct evidence on granular details and peripheral points, and to rebut fanciful speculations. It would increase demands on witnesses, because hundreds of witnesses will be needed to satisfy the types of requirements enumerated in Gbagbo. But the bigger problem is that the Cartesian approach can only produce acquittals in complex cases. This is because it hypersceptically and atomistically deconstructs each item of evidence; it is preoccupied with certainty and speculations; it rigidly applies analytical categories at the expense of substance; it fails to apply standard tools of reasoning to connect dots; it requires a granularity and tidiness that reality cannot furnish; and it is it is stymied by the minor discrepancies that inevitably arise between different accounts. There is little point reforming investigations or other procedures if these types of standards come to prevail.
For ICC observers, I realize that the mainstream view is to assume an inadequate investigation and blame the OTP. However, I would encourage independent and critical thinking. Look at the evidence in the majority and dissenting decisions, and ask yourself if there is other evidence you would have collected. Where the majority decision indicates a desire for more evidence, ask yourself whether the requests reflect existing standards, or do they ask for duplicative, tangential, irrelevant, or impossible-to-obtain evidence. This might lead to more nuanced diagnoses and thus more effective prescriptions for reforms.
For States Parties concerned about budget and efficiency, it is worth pondering the impacts of any novel evidentiary expectations on the cost and duration of investigations and trials.
As for the Gbagbo case, in my view, there are multiple significant errors of law in the Gbabgo majority’s approach to evidence that would easily warrant a new trial under national or international standards. However, the Appeals Chamber is sometimes understandably reluctant to order a new trial, because trial proceedings are so long and complex. It would occupy years more of the life of Laurent Gbagbo, and require a hundred witnesses travelling and testifying again, with many attendant complications. However, an irony is that a rise in Cartesian judging will only compound such problems, because trials will become much longer and more laborious.
In my view, the Court’s jurisprudence should push affirmatively for evidentiary standards closer to those developed by national and international systems in light of long practice. The more typical standards are practical, holistic, even-handed, they draw on experience, and they make warranted inferences from evidence. They also take into account feasibility (whether a type of evidence can even be collected) and procedural economy (whether ‘best’ evidence is always required). An invigorated comparative conversation about international criminal evidence law has never been more essential.