The Other Poisoned Chalice: Unprecedented Evidentiary Standards in the Gbagbo Case? (Part 2)

Written by

My aim in this three-part series is to start a conversation about unusual and problematic evidentiary standards emerging at the ICC.  These standards flow from a commendable impulse to uphold the highest standards, but they entail an unprecedented and unattainable exactitude. In my view, if these standards take hold, they will result in the repeated crashing of complex cases, making them especially poorly suited for precisely the types of cases the ICC is mandated to deal with. 

In my view, the more common and appropriate approach, seen in national and international practice, is even-handed, holistic, experiential and practical.  The experiential approach draws on human experience.  It employs sound methods of reasoning, such as triangulation, extrapolation, interpolation, and inference to best explanation, and thus it is can work judiciously with patterns and inferences.  It is also practical: it bears in mind feasibility and procedural economy.

For brevity, I will call the alternative, emerging approach the “Cartesian” approach. I introduced its features in part 1 of this series, such as its hyperscepticality, atomism, and fixation with certainty and speculative doubts.  In this post I will give some additional examples of problematic evidentiary approaches as seen in the Gbagbo trial decision.  As the judgments are over 1300 pages, I am only able to outline some of the concerns and some examples in the most general and cursory.  My hope is to trigger an invigorated discussion of international criminal evidence law.

Pointillistic Corroboration

The Gbagbo majority adopts a narrow and novel conception of corroboration.  I will call it the “pointillistic” conception, because it treats the evidence like a pointillist painting, with evidence being considered corroborative only if it confirms the same dot.  The majority writes, “corroboration only occurs when two pieces of evidence independently confirm the same fact” (§47) and that “corroboration… is proposition-specific. What connects different items of evidence is that they are relevant to the same fact.” (§48). 

At first blush, this sounds correct: evidence has to be about same fact to be corroborative. 

However, there are many ways that evidence can corroborate without pertaining to the same fact. 

 For example, if 60 women say Bill Cosby drugged them and then raped them, those are 60 different incidents, and hence different “facts”.  And yet, on a standard (experiential) approach, rather than looking pointillistically at the dots, we can see that that the testimonies corroborate a pattern of behaviour. 

Similarly, when different people report that a police force or armed group attacked them in different incidents, this is corroborative.  On an experiential account, we don’t just see discrete events.  We use standard tools of reasoning, such as ‘inference to the best explanation’.  Applying human experience, we explicitly or implicitly consider the main hypotheses: for example, that the witnesses are all lying, or the organization may have instructions or at least a culture encouraging such crimes.  The witness accounts corroborate each other, because they attest to the same pattern, strengthening the second hypothesis and making less likely the first.

The Gbagbo majority, unfortunately immersed in the pointillistic conception, casts caustic doubts on the idea that different incidents could corroborate each other, because they are ‘discrete events’ involving ‘different perpetrators’ (§1388, Tarfusser at §76).  However, in a crime against humanity case, the Prosecution is generally trying to show that crimes are not just acts of isolated rogue actors, but rather something symptomatic of the organization.  So that the fact that the crimes are committed by different perpetrators within the organization doesn’t weaken the argument, it strengthens it.  The majority position is almost perfectly backwards to what law and logic require. 

The problem with the pointillistic conception is that it examines the canvass from far too close: it fixates on the dots in too fine-grained a manner.  Assessing evidence is not an acontextual, mechanical, check-box process.  The judge’s job is to bring to bear wisdom, experience, and sound tools of reasoning.   This includes the ability to draw warranted inferences, to ascertain what happened.  Where one is trying to prove a pattern, or an organizational culture, or a common plan, then events from different incidents and involving different persons can in fact corroborate each other; indeed, ‘triangulation’ is often the only way to prove such ephemeral things. 

The pointillistic conception permeates and undermines the Gbagbo majority judgement, as it was left unequipped to assess corroboration of patterns of conduct, which was the core of the case.

Formalism over substance (eg hearsay)

Another problem with the Cartesian approach, embodied in the Gbagbo majority decision, is that it applies analytical categories with a zeal and rigidity not seen in national systems.  Again, the impulse is commendable – to ensure the quality of the evidence.  However, when applied with excessive austerity, it loses sight of substance and context, the purpose of the activity (to ascertain truth), and the need for feasibility and procedural economy.

This tendency is best illustrated by examples of the majority’s approach to evidence that it labels and dismisses as “anonymous hearsay”.  I focus on “anonymous hearsay”, because most readers will initially be most inclined to agree with a firm exclusionary approach for this category.

The Gbagbo majority declares that “no probative value can be ascribed” to anonymous hearsay (§43).  The reason for this is, at first glance, convincing: “no responsible adjudicator can base factual findings on evidence without having good reasons to accept that the source of the information is sufficiently trustworthy. In the case of anonymous hearsay, this is simply impossible because the source of the information is unknown and can therefore, by definition, not be evaluated” (§43).

Nonetheless, I believe that the majority’s approach is both legally and epistemologically unsound.  Anonymous hearsay can be generated in circumstances that can make it reliable and valuable, and in such instances it is routinely used in national courts and in daily life. 

Suppose you are in a store, and you see a child about to open and drink from a bottle marked “poison”.  Should you be concerned?  That label is anonymous hearsay. You don’t know which factory worker affixed that label.  You cannot assess his or her credibility.  Thus, on the majority’s analysis, you should give the “poison” warning no probative value at all.  You should go about your business completely unconcerned.  By contrast, on any approach that considers human experience, you would regard the label as probative (and usually highly probative).  So, it turns out we don’t always need to assess the credibility of the utterer; often the circumstances can furnish sufficient reliability.  This is the problem with the Cartesian approach: it gets too lost in its idealized analytical constructs, and loses touch with experience and context. 

The Gbagbo majority repeatedly dismisses UN reports, NGO reports, and media reports as “anonymous hearsay” (eg. §189, 202, 1102, 1151, 1152, 1321, 1801).  But as Judge Carbuccia rightly noted in dissent, those reports were only used to provide corroboration of direct evidence, and to establish surrounding context (eg. patterns, ‘widespread’, etc.).  Both in law and in common sense, a convergence of reports by diverse and independent credible actors is probative information.  Such reports are especially reliable where their creators testify as to their methodology; for example as in the Gbagbo case where the UN explained its methodology of requiring two independent sources per claim (§ 1401). My preliminary research indicates that national courts routinely use UN, NGO and media reports for corroboration and context.  National courts have specifically held UN and NGO reports to be credible and reliable.  Such reports strengthen our confidence in the overall pattern of events, while more specific evidence zooms in on particular incidents and responsibilities.  The combination of overview evidence and the specific evidence increases our confidence.

Here is another example of “anonymous hearsay”.  In numerous instances, the UN received large numbers of panicked emergency calls, with people reporting that the police were shooting at people and killing them, or shooting mortars at civilians, or that pro-Gbagbo forces were burning people alive.  The majority discredits these calls as “anonymous hearsay” (eg. §1315, 1322, 1360-62, 1366, 1620-21).  However, this is evidence that even in the common law, despite its famous aversion to hearsay, would be admissible (spontaneous excited utterances, res gestae).  In a standard approach, we do not just mechanically apply the category, we also consider substance and context.  Such an influx of calls is a strong indication that such crimes were occurring.  The large number of different callers provides an indicia of reliability, as does the corroboration from eye-witnesses and video evidence.

Authentication and Cartesian certainty

One of the most-quoted criticisms in the majority judgment is that “a majority of the documentary exhibits submitted by the Prosecutor in this case would not pass even the most rudimentary admissibility tests in many domestic systems” (§36).  The line has been repeated at face value as demonstrating the frailties of the Prosecution case.  But is the complaint accurate?  What tests would the documents fail, and why?

The documentary evidence included documents collected from police archives, bearing signatures of their authors.  The majority’s difficulty with authentication was that documents still might not be genuine (§33).  And it is true, as a purely hypothetical supposition: even though they look like government documents and were found in the archive for government documents, they might be clever and elaborate forgeries planted in a government office.  But this is a standard of Cartesian certainty.  I do not think that is the common standard in national systems, nor would it be a very practical standard.

The majority cites one witness, who would not confirm that he had signed a document that bore his signature, as a reason to doubt the entire archive (§33).  This is an unorthodox standard.  A more common (experiential) approach would bear in mind that a witness refusing to confirm his signature when confronted with an inconvenient or genuinely forgotten document is an entirely commonplace occurrence.  The normal reaction to one recalcitrant or cautious witness is not to doubt the authenticity of the entire archive.  Evidenceless speculations about large-scale forgery would be offset by briefly considering the implausibility of such an elaborate operation in the prevailing conditions.

Other governmental documents were seized from the presidential residence. Ivorian authorities informed the investigators that the palace “had been left untouched since the arrest of Laurent and Simone Gbagbo” (§35).  Nonetheless, the majority objected that “no evidence has been presented by the Prosecutor showing that no one entered the Presidential Residence in the period between Mr Gbagbo’s arrest and the Prosecution’s site visit” (§35).  Thus, according the majority “it is far from certain” that the documents were not tampered with (§35). 

Notice that the majority requires proof of an absence (i.e. a negative universal), which is impossible to satisfy.  The best available evidence was that the residence had been “untouched”.  More importantly, the prospect that an unknown person slipped in and tampered with the documents is entirely hypothetical.  This is a speculative doubt.  Furthermore, on a standard approach, drawing on context and experience, we would consider: if an imagined intruder had snuck in to plant false evidence against Gbagbo, wouldn’t he or she have planted something warranting the effort?  The fact that the documents were fairly quotidian helps dispel the purely speculative hypothetical about tampering.

In addition, the majority repeatedly comments adversely on documents that are not “signed, stamped and dated” (§100, 537, 543, 574, 668, 678, 681, 685, 745, 937, 945, 1084, 1238).  Of course it is helpful when a document is signed, stamped and dated – that gives more information about the document.  But that is not reasonable as a baseline expectation.  The vast majority of genuine documents generated by human beings in this world are not signed, stamped and dated.  The fact that the judgement remarks critically on each document failing to meet all three of these conditions is an example of how the Cartesian approach focuses on its idealized pre-requisites for certainty, without an adequate ‘reality check’ as to the soundness of the expectations for assessing real-world documents.

Fastidiousness: fixation with minor discrepancies

In earlier days, Tribunal judges were understandably criticized for neglecting major inconsistencies in evidence (see eg Nancy Combs).  Unfortunately, the Gbagbo majority falls into a major over-correction, with wholesale rejection of evidence because of minor and commonplace inconsistencies. 

Empirical studies show that eye-witness recollections vary, especially after the passage of time and especially in relation to traumatic events.  Whenever we try to reconstruct complex events, there will always be minor discrepancies.  Holistic, experiential judging does not discard evidence as unreliable because of minor discrepancies.  Instead, it looks at the evidence with all of its frailties, using other evidence that corroborates an account, in order to assess whether the elements are satisfied beyond reasonable doubt.  (In fact, it is when several witnesses recall events with perfect consistency that one should really start to worry about reliability!)

There are countless examples in the judgment of the excessive preoccupation with details.    A preponderance of evidence (eyewitnesses, videos, experts) about a tank opening fire on unarmed civilians was undermined by minor discrepancies (eg the number of shots fired, or whether witnesses saw two or five vehicles) (§1773-87).  The assessment of insider evidence about a unit firing mortar shells floundered due to inconsistencies about who set up the mortars and how many pieces there were (§1811).  Insider witnesses differed slightly on how the police and army coordinated, so the majority declared itself unable to draw any conclusions at all (§1174-78). A particular fascinating reading is the intricate and almost obsessive exposition on minor mistakes in the logbook of the presidential residence (§304-315).  On an experiential approach, we are not remotely surprised if an artifact created by different guards contains mistakes and duplications.  Nonetheless if the log shows, for example 105 meetings (§315), then we can allow for some mistakes and still have, beyond reasonable doubt, a lot of meetings.  On the Cartesian approach, however, if something does not provide perfect certainty, it is “unreliable”.

Fastidiousness: insistence on inappropriate granularity

Another feature of Cartesian judging is insistence on a level of detail and granularity that is legally unnecessary and epistemologically unattainable.  Again, the impulse is well-intentioned: it flows from a belief that rigour requires such exacting detail.  In effect, the Cartesian judge is asking, “How can I conclude X unless you prove to me details X1, X2 and X3?”.  And, having proof of X1, X2 and X3 can indeed help us to prove X.  The error arises when one starts to think that proof of X1, X2 and X3 are legal requirements.  They are not.  If the element is X, then we only need to conclude X beyond reasonable doubt, and we do not need to be sure of details X1, X2 or X3.

Here is an example of the ‘granularity’ tendency.  The Prosecution argued that the police and other pro-Gbagbo forces attacked civilians perceived to be sympathetic to Ouatarra.  However, the majority craved even greater precision and detail.  The majority complains that “Prosecutor does not point to any uniformly applicable criterion that would qualify an individual and/or group as Mr Gbagbo’s political opponent by definition” (§172).  The majority wanted to know “with specificity what criteria the perpetrators were using in determining their perception of their victims” (§183).  Accordingly, the Prosecution gave additional granularity, specifying five subgroups in the targeted group (§173).  In response, the majority complains that this “adds confusion to the prosecution narrative.”  {§175).   In addition, the majority simply descended to a new level of granularity, complaining that “the Prosecutor fails to advance clear criteria to delineate the different constituent groups” (§1397).

The problem here is demanding a clinical clarity and granularity that reality might not provide.  Mass atrocity will entail evolving plans and reactions, and different ringleaders and perpetrators may have slightly diverging aims and understandings.  Even if we had an omniscient perspective – i.e. we didn’t have to worry about evidence – it is highly unlikely that the requested level of analytical tidiness exists even in the minds of ringleaders and perpetrators. 

The Nazis killed not only Jewish people; they also killed political opponents, dissidents, disabled people, homosexuals, and Roma.  The Nuremberg judges did not complain that this assortment of target groups “confused the prosecution narrative”. It’s just an unfortunate fact about what the Nazis did.  A criminal court has to deal with reality.  Because of its obsession with tidiness and precision, the Cartesian approach expects precisely-formulated and consistently-applied criteria.  But mass atrocities will probably never entail the “clear”, “specific” and “uniformly applicable” criteria required by the Gbagbo majority.   


In this post I have introduced some more features of the Cartesian approach: pointillistic corroboration, analytical formalities over substance, unrealistic authentication, balking at commonplace minor inconsistencies, and craving for granularity and clinical clarity. At the root of all of this is a well-intentioned but highly fastidious way of thinking.  Criminal law is supposed to operate, not like a thought experiment, but at a ‘human’ level.  In my view, the Cartesian approach departs from national and international practice, it is epistemologically unsound, and it is problematic for the viability of the Court and the fulfilment of its mandate.  In the third and final post, I will provide further illustrations and conclude with implications.


Print Friendly, PDF & Email

Leave a Comment

Comments for this post are closed


Darryl says

November 6, 2019

I should add a comment to answer questions/critiques from Dov Jacobs:

- In the "poison label" example, notice that I am talking about the basic threshold of whether the evidence is "probative". Not whether it proves the fact beyond reasonable doubt. Those are two different thresholds. So, in that segment I am not lowering, or even discussing, the reasonable doubt standard. I am talking about whether the evidence can be admitted and given any weight at all.

- Dov gives a counter-example of a "beware of dog" sign. This entirely misses my point that sometimes the CIRCUMSTANCES can generate reliability. For example, the poison label is a routine utterance made in the course of duty.

- I think that "outsiders" to a trial are allowed to discuss the judgement, including on the law of evidence. I am discussing the appropriate legal methodology for assessing evidence, which is a legitimate and important topic.

- I agree that judges should be 'sceptical', but my posts are about 'hyper-scepticism'.

-Urging adherence national and international practice does not mean that a person does not care about the 'reasonable doubt' standard.