On June 8, 2018, in a surprising turn, the Appeals Chamber of the International Criminal Court (ICC) reversed the conviction of Jean-Pierre Bemba Gombo and acquitted him of crimes against humanity and war crimes. The four separate opinions, raising questions about Pre-Trial and Trial Chamber procedures, the standard of Appellate Chamber review, and the scope of command responsibility, have revealed sharp disagreements between ICC judges and created considerable confusion over the state of ICC law and procedure.

Bemba, a Congolese national, was convicted of crimes against humanity and war crimes committed by Movement for the Liberation of Congo (MLC) troops in the Central African Republic (CAR) by an ICC Trial Chamber on March 21, 2016Footnote ¹ and sentenced to eighteen years imprisonment.Footnote ² The Trial Chamber found that Bemba was the president of the MLC, the leader of its political branch, and held the military rank of divisional general (para. 384). As such, he “held ultimate authority over sanctioning, arresting, and dismissing senior political leaders and military officers, as well as soldiers” (para. 403). Although “predominantly based” in the Democratic Republic of the Congo and “therefore remote from the operations on the ground,” the Trial Chamber found that in addition to his formal authority, he “visited the CAR on a number of occasions” and regularly communicated with his forces (paras. 706–08). The Trial Chamber concluded that he had “direct knowledge” of MLC crimes (para. 710) and failed to take “all necessary and reasonable measures to prevent or repress [them]” (p. 350).

The Bemba trial judgment was noteworthy and important.Footnote ³ It was the first case to find an accused guilty of command responsibility under Article 28 of the Rome Statute, as well as the first ICC conviction for sexual violence. It was not a “particularly controversial” decision,Footnote ⁴ and received generally positive scholarly commentary.Footnote ⁵ The three trial judges were unanimous in their assessment of Bemba's culpability, although two—Judges Sylvia Steiner and Kuniko Ozaki—raised questions regarding causation and whether a commander's failure to exercise control was a necessary precondition to holding him or her responsible for the crimes’ commission.

As the ICC's pre-trial procedure requires, the case had been committed to trial by a unanimous “Confirmation Decision” rendered by Pre-Trial Chamber III in 2009, finding “substantial grounds to believe that Mr Jean-Pierre Bemba … effectively acted as a military commander … [for] crimes against humanity… and … war crimes…” (para. 446).Footnote ⁶ The Confirmation Decision was based upon a “Document Containing the Charges” filed by the prosecutor, which was amended twice and ultimately finalized as a “Corrected Revised Second Amended Document Containing the Charges” with the Trial Chamber on October 8, 2010. The trial began in November 2010.

On June 8, 2018, the ICC Appeals Chamber reversed and acquitted Bemba in a fractured 3–2 decision setting out the judges’ views in four separate opinions. The Majority found that Bemba's conviction exceeded the facts and circumstances described in the charges brought against himFootnote ⁷ and that the Trial Chamber paid “insufficient attention to the fact that the MLC forces” under his command “were operating in a foreign country with the attendant difficulties on Mr. Bemba's ability, as a remote commander, to take measures” to control their actions (para. 68). Two members of the Majority, Judges Morrison and Van Den Wyngaert, appended a Separate Opinion declining to permit a retrial.Footnote ⁸ Judge Eboe-Osuji, president of the ICC, joined the Majority but argued in his Concurring Separate Opinion that he would have permitted a retrial, but having been unable to persuade his colleagues to join him, felt obliged to acquit.Footnote ⁹ Judges Monagang and Hofmański penned a lengthy Dissent.Footnote ¹⁰

The Majority found that if an accused “identif[ies] sources of doubt about the accuracy of the trial chamber's findings … the Appeals Chamber [will] independently review the trial chamber's reasoning on the basis of the evidence that was available to it” (para. 2). It stated that “[m]ere preferences or personal impressions of the appellate judges are insufficient to upset the findings of a trial chamber” but that if a “reasonable and objective person can articulate serious doubts about the accuracy of a given finding,” it is a “strong indication,” that the trial chamber may not have respected the standard of proof requiring findings to be made “beyond [a] reasonable doubt” (paras. 44–45).

Because Bemba was charged with command responsibility under Article 28 of the Rome Statute, the Majority examined the Trial Chamber's findings as to the crimes committed by his troops for which he was allegedly responsible, holding that Bemba could only be convicted “of the specific criminal acts of murder, rape and pillage that the Trial Chamber found to be established beyond a reasonable doubt and which were … recalled in the concluding sections of the Conviction Decision in relation to each crime” (para. 104). The Appeals Chamber also criticized the Confirmation Decision for containing language that was “too broad” (para. 110). It noted that there were individual acts alleged in the charging document upon which the case could proceed to trial (para. 111), but that Bemba's conviction was limited to acts referenced directly in the Confirmation Decision itself (paras. 115–16). Other crimes listed in the charging document could not sustain the conviction but could be “taken into account for the finding regarding the contextual element of crimes against humanity, which operates at a higher level of abstraction” (para. 117).

The Majority further found that Bemba could only be held responsible under Article 28 for failing to undertake measures that were “at his … disposal in the circumstances at the time” (para. 168). These measures must have been “specifically” identified by the Trial Chamber and shown to have been measures “which a reasonably diligent commander in comparable circumstances would have taken” (para. 170). The Majority found that he did not have adequate notice of the specific measures that the Trial Chamber found to be required of him (such as redeployment of his troops) (para. 188) and, overall, that the Trial Chamber had failed to “fully appreciate the limitations that Mr Bemba would have faced in investigating and prosecuting crimes as a remote commander sending troops to a foreign country” (para. 191)

* * * *

Bemba's acquittal was surprising.Footnote ¹¹ Some have suggested it was the prosecutor's fault.Footnote ¹² This seemed to be the view of Judges Morrison and Van den Wyngaert, who mused that Bemba might have been held criminally liable had the prosecutor brought different charges or found stronger evidence (para. 78). A careful study of the four opinions, however, reveals that the acquittal resulted from the widely disparate views of the five members of the Appeals Chamber as to their role, the ICC trial and pre-trial process, and the applicable legal standards.Footnote ¹³ Judges Morrison and Van Den Wyngaert said as much, noting that the inability to achieve consensus “appeare[d] to be a fundamental difference in the way we look at our mandates as international judges” (sep. op., para. 4). The analysis that follows addresses only a few highlights of the decision: the controversy over the new standard of review employed by the Majority; the question of command responsibility; the policy element for crimes against humanity; and the vexing problem of precedent and separate opinions in ICC judgments.Footnote ¹⁴

As for the standard of review, earlier ICC cases, including the Appeals Chamber's first judgment in Lubanga, held that the deciding factor in reviewing facts on appeal is whether a reasonable trial chamber could have been satisfied beyond a reasonable doubt as to the finding in question.Footnote ¹⁵ Although the Bemba Majority cited Lubanga, it found that it was entitled to “interfere with the factual findings of the first-instance chamber whenever the failure to interfere may occasion a miscarriage of justice” (para. 40). The president's Separate Concurring Opinion further suggests that appellate deference to factual findings is “not a matter of law” but a question of “judicial policy” (para. 46). He concluded that appellate review at the ICC should not follow the practice of the ad hoc international criminal tribunals but should be “reformulated” based upon his interpretation of Article 83(1) of the Rome Statute (para. 8).Footnote ¹⁶

The president's view is controversial. Article 21 sets forth a specific methodology for Rome Statute interpretation, requiring judges to refer first to the text of the Statute, then to relevant treaties and customary international law, and finally to general principlesFootnote ¹⁷ (pp. 760–62). The president's opinion commences with a philosophical argument, navigates to the text of the Statute, and then considers commonwealth cases in support of his proposed interpretation. Yet reliance on national cases is disfavored in international criminal adjudication,Footnote ¹⁸ and the specific cases cited actually support a classic standard of deference to lower court findings on questions of fact, not de novo appellate review.Footnote ¹⁹

As a policy matter, the value of the Appeals Chamber engaging in de novo review is not obvious, even if, as the president suggests “with the utmost humility that the accumulated weight of (sic) majority's experience in international criminal law alone is no less than 55 years” (concurring sep. op., para. 9, n. 1). As other Appeals Chambers have held, “Trial Chambers are best placed to hear, assess and weigh the evidence, including witness testimonies, presented at trial.”Footnote ²⁰ In the Bemba case in particular, given the proven allegations of witness tampering that contaminated much of the testimony offered by the defense team during the trial,Footnote ²¹ the Appeals Chamber Majority may have unwittingly accepted a narrative that was later proven to be false because, unlike the Trial Chamber, it did not have the opportunity to hear and assess all the evidence.Footnote ²² Finally, the “desiderata of consistency, stability and predictability, which underlie a responsible legal system,” suggest that a court should not depart from prior decisions except with circumspection, so as not to “unnecessarily impair[] the authority of its decisions.”Footnote ²³

The proper interpretation of Article 28 (command responsibility) was heavily briefed before the Appeals Chamber. Judges Van Den Wyngaert and Morrison wrote that the “effective control” test, one of the elements required in the application of Article 28, is limited to platoon or section commanders, rather than higher-level commanders, who are “entitled to rely on lower level commanders to keep their troops in check and to deal with deviant behavior” (sep. op., para. 33) The Dissent disagreed. The president, in his concurring opinion, helpfully discussed the importance of command responsibility in the context of international crimes and the purposes of the Rome Statute. Because the four opinions are contradictory, they decide very little. Two members of the Appeals Chamber (Monagang and Hofmanski) believe that causation is required under Article 28. Judges Morrison and Van den Wyngaert disagreed. President Eboe-Osuji notes that causation is a “perennial thorn in the side of criminal law” (para. 156) and concluded that the defendant's contribution must be “more than negligible” (para. 166).

The Appeals Chamber also split 3–2 on the “necessary and reasonable measures” that a commander is required to take under Article 28. The Majority argued for an assessment “of what measures were at his or her disposal in the circumstances at the time” (para. 168), and criticized the Trial Chamber extensively, identifying seven specific errors in its assessment. The Dissent argued that the Majority's analysis represented selective and unwarranted de novo review, and itself reviewed the evidence more comprehensively. The president found that one measure Bemba could have taken to stop the crimes was to withdraw MLC troops from the CAR, agreeing with the prosecution that the “obligation to withdraw is … entirely a matter of common sense” (concurring sep. op., para. 280). Yet he declined to find Bemba legally responsible for failing to do so (concurring sep. op., para. 282).

The Majority's further suggestion that Bemba's status as a “remote commander” “sending troops to a foreign country” limits his responsibility is troubling (para. 191). This unprecedented statement—made without reference to supporting legal authority—shelters all commanders sending troops abroad. Yet those commanders should arguably be required to exercise a higher level of due diligence and supervision than the Majority seems to require precisely because of the risks involved. Given that today's military commanders have almost immediate access to their forces through modern communication methods, there seems little practical reason for concern with imposing such a standard. Although joining the Majority, the president seemed to disagree with this finding, holding that the “geographic remoteness of a commander” would not necessarily insulate him from criminal responsibility (concurring sep. op., para. 258).

The Appeals Chamber requested briefing on the “state or organizational policy” element of Article 7 (defining crimes against humanity), but the Majority opinion is silent on that question. Defense lawyers had argued that the element imposes a high threshold, and suggested, like the late Judge Hans-Peter Kaul, that “state action” is part of the policy element (paras. 51–53).Footnote ²⁴ The president's opinion addressed and rejected these arguments, noting that the Court's prior decisions had already settled the question. He observed that the Rome Statute must not be “compromised in its ability to protect humanity against the gross atrocities that fall within the class of crimes over which the Court enjoys jurisdiction” (concurring sep. op., para. 286). The Separate Opinion, however, held (without discussion or authority) that Article 7(2)(a) requires the prosecutor to prove beyond a reasonable doubt each individual instance of criminal conduct alleged to be part of the “course of conduct” (sep. op., para. 66). The Dissent disagreed, so the judgment does not decide these questions.

The Separate Opinion's findings on the quantum of proof and appropriate assessment of the context elements are puzzling. Context elements are jurisdictional in nature.Footnote ²⁵ Whether a conflict is international or non-international or whether there is an “attack directed against a civilian population” separates ordinary murders and other acts of violence from international crimes. Under Article 7(2)(a) of the Rome Statute, the prosecutor must prove the “course of conduct” that constitutes the “attack” directed against the civilian population.Footnote ²⁶ The reference to “acts referred to in paragraph 1” is a limitation on the kinds of acts that can be used to establish the course of conduct, not an invitation to the judges to require that every subsidiary act giving rise to the finding of the “course of conduct” be independently established beyond a reasonable doubt. If a defendant is accused of dispatching a train carrying civilians and deporting them to a concentration camp, that would represent a “course of conduct” involving the multiple commission of “acts” establishing one of the context elements; it would be impossible to establish each individual act of deportation in most situations, nor does the case law require the prosecutor to do so. Indeed, given the practical reality of mass atrocity prosecutions, it is unsurprising that the ad hoc tribunals have not adopted the Separate Opinion's interpretation of Article 7(2)(a), which seems grounded in neither the text of the Statute nor customary international law.

Likewise, as to Bemba's second ground of appeal (the conviction exceeded the charges), the Majority appears to be reading the Statute in an unprecedented manner. Bemba did not complain of a lack of notice of the charges against him; rather, he argued that only acts actually referred to in the Confirmation Decision could form part of the case against him (para. 99). At one point, the Majority appears to reject this complaint, finding that acts listed in the Document Containing the Charges were included within the scope of the trial (para. 113). But it nonetheless reversed course by finding that only the acts forming part of the case—that is, the charges specifically listed and proven beyond a reasonable doubt in the Trial Chamber's judgment (para. 102)—could form part of the Conviction Decision. The reasoning of this section of the Majority Opinion is somewhat confusing, and the dissenting judges contested it vigorously. As the Majority notes, the Amended Document Containing the Charges listed “a number of alleged criminal acts of murder, rape and pillaging,” but indicated that this list was not exhaustive (para. 75). The final Document expanded the list of criminal acts, while also using the word “including,” and the prosecutor relied upon additional acts during the trial proceedings to establish the crimes alleged to have been committed by Bemba's forces. The prosecutor, the Pre-Trial Chamber, and the Trial Chamber all took the view that “the scope of the charges was not limited to the individual incidents of killings, rapes and pillaging discussed in the Confirmation Decision, but extended to all such acts committed by MLC soldiers against CAR civilians on CAR territory from on or about 26 October 2002 to 15 March 2003,” assuming of course that Bemba had received actual notice in a timely fashion of the alleged crimes (para. 91). Bemba was accordingly convicted of “crimes of murder, rape and pillaging committed by MLC soldiers” during the specified period based upon the acts alleged to have been committed and proven by the prosecutor (para. 93).

In a new interpretation of Article 74(2) of the Statute, the Majority asserted that this view was too broad. The Majority complained that the Trial Chamber had not referenced “even an approximate number of the individual criminal acts of murder, rape and pillage,” and limited the scope of the conviction to the acts specially referenced by the Trial Chamber “recalled in the concluding sections of the Conviction Decision in relation to each crime” (paras. 103–04). As proposed by the defense, the Majority further held that the Trial Chamber could not have considered “criminal acts that the Prosecutor added after the Confirmation Decision was issued” (para. 115).

Five members of the Court (the dissenters and the Trial Chamber) disagreed, and would have sustained the conviction. This remains a serious interpretive split. The Dissent articulates the classic position: that it is within “the Prosecutor's discretion to formulate charges … at a broader level” and charges may be confirmed as such by the Pre-Trial Chamber (dissent, para. 39). Thus, the dissenters were content with the formulation and conviction, noting that “in the present case, the charges were delineated by temporal, geographical and other factual parameters; they were further specified by reference to the description of the advance through and withdrawal from the CAR of the MLC troops” (dissent, para. 39). Thus, taking into account the mode of criminal responsibility charged, as well as the remote position of the accused and the numbers of crimes alleged, the Dissent concluded that the “conviction did not exceed the facts and circumstances described in the charges” under Article 74(2) of the Statute (dissent, para. 22).

The result in this case is unsatisfying. The inability of the Appeals Chamber to achieve consensus means that the judgment decided very little as a matter of law yet considerably muddied ICC procedure and practice. The Separate Opinion observes that the issuance of dissents and separate opinions is common practice in international tribunals. Such is not always the case. The European Court of Justice was arguably successfulFootnote ²⁷ establishing its independence precisely because it issued only consensus opinions. While dissents can be helpful in clarifying the law, in Bemba the disparate views in the four separate opinions confuse the law and were released seriatim, with the president's opinion following the other two. This led to speculation as to why the opinions were not available at the same time and to an uncertain understanding of the judgment.

The Appeals Chamber's decision also highlights weaknesses in the ICC's trial process. Pre-Trial Chamber II had limited the case to one mode of liability, as was the practice at that time, rejecting the possibility that Bemba could be charged as both a superior under Article 28(a) and as a co-perpetrator under Article 25(3) (para. 342).Footnote ²⁸ It also found that it was appropriate, given the early stage of the proceedings, for the prosecutor to provide “not all but only sufficient evidence” and found unobjectionable the use of the phrase “include, but … not limited to” in the Document Containing the Charges, language now rejected by the Majority (para. 66). This 2009 decision committed the case to trial, but was essentially reversed by the Appeals Chamber nine years later, in a manner that seems inconsistent with the directive of the ICC Chamber's Practice Manual. The Manual underscores the limited nature of the Pre-Trial Confirmation process as ensuring a case has met the relevant legal and evidentiary thresholds to be committed to trial, but in no way represents a “mini-trial” of the charges leveled against the accused.Footnote ²⁹ No legal system functions well with this level of unpredictability, which is particularly damaging to a fledgling court.

Judges Morrison and Van Den Wyngaert voted to acquit rather than order a retrial because they worried their fellow judges would suffer from a “perverse incentive” to convict to justify Bemba's detention; moreover, they did not think it fair to reward the prosecutor with a “second chance” given the “serious problems” they detected in the first case (sep. op., para. 73). Yet the point of Bemba's acquittal was surely not to “punish” the prosecutor for misconstruing law that had not yet been clearly articulated by the judiciary, but to protect the rights of the accused, no matter the cost to the victims and the Court itself. Although acquittals are a normal and desirable feature of international criminal adjudication, this particular acquittal seems problematic. On June 18, 2018, Bemba returned to Belgium to rejoin his family. He was subsequently sentenced to one-year imprisonment and a EUR300,000 fine for offenses against the administration of justiceFootnote ³⁰ (pp. 50–51) resulting from his participation in a “vast and systematic … criminal enterprise” to taint the evidence in the case.Footnote ³¹ In such circumstances, even taking the majority's view of the proof adduced in the main case, it seems that ordering a retrial would have been both fair and appropriate.