From Fact-Finding to Fact-Proving at the ICJ: The Gambia v Myanmar and the Court’s Coming Reckoning on Evidence, Interventions, and Intent

 Khan Khalid Adnan is a Fellow of the Chartered Institute of Arbitrators (FCIArb), a Barrister in England and Wales, and an Advocate of the Supreme Court of Bangladesh. He currently serves as the Head of the Chamber at Khan Saifur Rahman & Associates, Dhaka, Bangladesh.]

Oral proceedings on the merits in The Gambia v Myanmar(12-29 January 2026) ended where genocide litigation before the International Court of Justice (ICJ) almost always ends: not with a dispute about atrocities as such, but with a dispute about how the Court is permitted to know what it knows. The Gambia framed the case as exceptional only in the thickness, corroboration, and institutional insulation of the evidentiary record. Myanmar framed it as a warning against outsourcing adjudication to non-judicial bodies, and against importing criminal rhetoric into a civil forum while disavowing the discipline of criminal proof.

The stakes are unusually high. The Court’s only clear finding that genocide occurred under the Genocide Convention remains Srebrenica in Bosnia and Herzegovina v Serbia and Montenegro, where the respondent State was held responsible only for failure to prevent and punish, not for commission (Bosnia Genocide Judgment, paras. 209, 297, 373). A merits judgment in The Gambia v Myanmar that attributes commission of genocide to a State would therefore mark a genuine jurisprudential rupture. That prospect should not be romanticised. The Court’s evidentiary architecture, especially the “fully conclusive” standard and the “only reasonable inference” approach (i.e., “the only inference that could reasonably be drawn”) to circumstantial proof of dolus specialis, is structurally designed to make genocide findings institutionally rare (Bosnia Genocide Judgment, paras. 209-210, 373; Croatia v Serbia, para. 148).

The January 2026 hearings exposed a deeper difficulty. The Court is now being asked to adjudicate allegations of genocide without the stabilising crutch that made Srebrenica comparatively manageable: prior criminal findings by an international tribunal. Whether the ICJ can adapt its evidentiary practice to this reality will shape not only this case, but the future justiciability of the Genocide Convention itself.

The UN Fact-Finding Mission: Evidence, Authority, and the Court’s Anxiety about Delegation

The UN Independent International Fact-Finding Mission on Myanmar (FFM) is the gravitational centre of the evidentiary dispute because it occupies an uneasy institutional position. Its mandate was investigative and preventive, not adjudicative. Yet the ICJ itself relied on FFM reporting at the provisional-measures stage, treating it as highly probative of ongoing serious violations without converting it into final judicial findings (Provisional Measures Order, paras. 55-56). That move now haunts the merits phase. The Court cannot plausibly pretend that the FFM is epistemically irrelevant, but it also cannot accept the Mission’s conclusions as a substitute for judicial fact-finding without collapsing its own standard-of-proof discipline.

The Gambia sought to normalise reliance on the FFM by embedding it within the Court’s established practice of according particular authority to certain UN materials when their preparation is careful, independent, and source-rich. Counsel expressly invoked the Court’s treatment of the UN Secretary-General’s Fall of Srebrenicareport, emphasising that the ICJ gave it “considerable authority” because of its comprehensive sources and institutional independence (CR (Oral Proceedings), pp. 16-17, para. 13) . The strategy is intelligible: if UN reporting could be relied upon to situate Srebrenica, it should not be dismissed simply because Myanmar refused cooperation.

Myanmar’s rejoinder was not that UN reports are inherently worthless, but that provenance alone confers no probative value. Counsel distilled this into a methodological filter: source, method, quality, and access. There is “nothing inherent” in UN or NGO status that gives weight “as such”; at most such reports are usable only insofar as they are probative and corroborated (CR (Oral Proceedings), p. 25, paras. 12-13). This move is familiar in ICJ practice, but it exposes a structural asymmetry. In mass atrocity contexts, the most “court-like” evidence often exists precisely because a State obstructs access.

That asymmetry surfaced repeatedly. The Gambia accused Myanmar of attempting to “disqualify” the FFM and IIMM (Independent Investigative Mechanism for Myanmar) record by invoking incompleteness caused by its own refusal to cooperate, while intervening States argued that UN reports can be especially credible because they emanate from a “disinterested witness” (CR (Oral Proceedings), pp. 21-22, paras. 25-27 (and fn. 32)).

The Court’s most coherent path lies between embrace and rejection. A principled approach would treat the FFM as a high-value repository of patterns, corroboration points, and investigative leads, while insisting that the “fully convinced” threshold must be satisfied through cumulative assessment: witness testimony tested in court, satellite imagery, contemporaneous documents, and adverse inferences drawn from obstruction. To do otherwise risks either delegating adjudication to the FFM or laundering denial through evidentiary formalism.

Myanmar pressed for the latter outcome in its sharpest form, arguing that because the FFM applied a lower standard of proof, “no reliance can be placed” on its findings when the ICJ must decide under a “fully conclusive” standard (CR (Oral Proceedings), p. 28, paras. 73-74). The argument is rhetorically powerful but conceptually unstable. The Court is not bound by the FFM’s inferential thresholds; it is bound only by its own. Different mandate does not entail categorical inadmissibility.

Article 63 Interventions and the Quiet Reconfiguration of Evidence Law

The doctrinal novelty of this case lies not merely in the number of Article 63 interventions, but in what the Court’s admissibility orders signal about the content of permissible intervention. The Order of 3 July 2024 emphasised that Article II of the Genocide Convention defines genocide through acts plus a special intent element, and framed that definition as a matter of construction (Order of 3 July 2024, para. 43). The Order of 25 July 2025 went further, stating explicitly that the definition of specific intent in Article II is relevant to the construction of the Convention (Order of 25 July 2025, para. 59). At the same time, the Court reiterated the familiar boundary: interventions are limited to construction; observations going to facts or application are not to be considered (Order of 25 July 2025, paras. 58-60).

This matters because disputes about what constitutes adequate evidence of dolus specialis, and about the inferential method the Court is willing to employ, can now be reframed as interpretive rather than merely procedural. Judge Cleveland’s declaration crystallised the point by foregrounding the parties’ long-standing disagreement over the evidentiary threshold for specific intent, including the applicable standard of proof (Declaration appended to the Order of 25 July 2025, para. 4; see here for more details).

The implications are unsettling. First, Article 63 intervention risks becoming a vehicle for collective evidentiary calibration. Interveners cannot submit case-specific facts, but they can exert interpretive pressure on what types of evidence are deemed capable of establishing intent. The January 2026 record itself reflects this dynamic, with interveners emphasising the special probative value of UN fact-finding for drawing inferences about dolus specialis (CR (Oral Proceedings), p. 21, para. 25 (and fn. 32)).

Second, mass interventions raise equality-of-arms concerns that the Court has historically addressed managerially rather than doctrinally (Whaling in the Antarctic; Order of 6 February 2013, para. 17; Declaration of Judge Owada, para. 5). In genocide litigation, where intent straddles law and fact, the boundary between interpretation and proof is unusually porous. Coordinated interpretive pressure can begin to resemble collaborative litigation strategy, even if formally confined to construction.

Intent, Inference, and the Architecture of Restraint

The hearings made unusually explicit what ICJ genocide jurisprudence often leaves strategically ambiguous: the dispute is not only about the standard of proof, but about what that standard does to inferential reasoning.

Both parties accepted that genocide allegations trigger a heightened standard, commonly expressed as requiring the Court to be “fully convinced” (Bosnia GenocideJudgment, paras. 209-210). The disagreement concerned whether this standard is functionally equivalent to “beyond reasonable doubt,” and whether it mandates the “only reasonable inference” (ORI) method for proving dolus specialis.

The Gambia, through Professor Sands, attempted a disciplined separation. “Fully convinced,” he argued, is the Court’s international standard; “beyond reasonable doubt” is a common-law criminal formula that the Court has never adopted and should not import (CR (Oral Proceedings), pp. 23-24, paras. 41-46). The position was not semantic but structural: importing criminal standards would transform the ICJ into a surrogate criminal tribunal.

Myanmar’s response was to weaponise the Court’s own language. Counsel argued that “fully convinced” is comparable to “beyond reasonable doubt,” and that The Gambia “tries to avoid” the criminal formula “like the plague” because it knows it cannot meet it (CR (Oral Proceedings), pp. 13-14, paras. 9-13). The rhetorical aim was to lock the Court into the strictest possible evidentiary posture without requiring an explicit doctrinal admission that the ICJ applies criminal proof standards.

Both sides converged on the ORI method. Myanmar insisted that genocidal intent must be the only reasonable inference from the established pattern of conduct, and that the presence of any alternative reasonable explanation defeats the claim (CR (Oral Proceedings), pp. 18-19, paras. 36-39). The Gambia accepted the same hook, but argued that the cumulative pattern of conduct admits no reasonable inference other than destruction of the Rohingya in part (CR (Oral Proceedings), p. 23, para. 43).

The ORI method is often presented as neutral logic. It is not. It structurally privileges respondent States in atrocity cases because alternative narratives are almost always available: counter-terrorism, loss of control, collateral damage, or mixed motives. Myanmar’s narrative architecture relied heavily on such alternatives, emphasising uneven geography, short temporal windows, and alleged counter-insurgency objectives as inconsistent with an exterminatory plan (CR (Oral Proceedings), pp. 69-70, paras. 108-111).

If the Court treats ORI as a near-automatic corollary of “fully convinced,” the case may hinge less on the horror of the record than on whether judges find a counter-insurgency narrative “reasonable.” That is an inherently normative judicial move, even when presented as evidentiary restraint.

Beyond the FFM: The Broader Evidentiary Ecology

Four additional evidentiary issues surfaced in ways that may shape the judgment. First, witness testimony. The Gambia emphasised that the Bench heard witnesses directly, seeking to convert empathy into epistemic legitimacy under the “fully convinced” standard (CR (Oral Proceedings), p. 23, para. 42). Myanmar urged scepticism, particularly regarding prepared and anonymous testimony (CR (Oral Proceedings), pp. 28-30, paras. 23-33).

Second, the IIMM’s limited production. Myanmar argued that the small number of witness statements annexed by The Gambia undermines reliance on earlier FFM and NGO material (CR (Oral Proceedings), pp. 55-56, paras. 15-18). If accepted, this argument would reward obstruction by allowing evidentiary scarcity to be weaponised.

Third, internal State documents. The Gambia attacked Myanmar’s reliance on military reports as “self-serving” and insisted on independent corroboration (CR (Oral Proceedings), pp. 36-37, para. 40). Uncritical reliance on curated internal archives risks normalising a State’s ability to litigate atrocity claims through its own paperwork.

Fourth, intent-signalling harm. The Gambia argued that targeting Rohingya children has singular importance for genocidal intent because there is “no other conceivable explanation” for such conduct at scale (CR (Oral Proceedings), pp. 28-30, paras. 9-13). Whether the Court treats such indicators as legally relevant to dolus specialiswill shape the future of inferential genocide analysis.

Conclusion: Will the Court Allow Genocide to Be Proved?

The central question after January 2026 is not whether the Rohingya suffered mass atrocity. The record is saturated with violence, destruction, displacement, and discriminatory structures. Even Myanmar’s litigation strategy tacitly concedes this by shifting the fight to inference and proof.

The real question is whether the ICJ will maintain an evidentiary architecture that renders genocidal intent nearly impossible to establish without prior criminal adjudication, or whether it will adapt its approach to fact-finding missions, corroboration, and inference so that State responsibility for genocide is not doctrinally confined to the shadow of criminal tribunals.

If the Court wishes to avoid being the first to attribute commission of genocide to a State, it already has the tools: marginalise the FFM, enforce ORI at maximal strictness, and treat alternative narratives as inherently reasonable. If it wishes the Genocide Convention to function as a judicial instrument rather than a symbolic charter, it must do something harder. It must explain, with methodological clarity, how a court can become “fully convinced” of genocidal intent in a world where States destroy evidence, deny access, and litigate through denial.

Either way, The Gambia v Myanmar will not merely apply existing ICJ genocide jurisprudence. It will expose whether that jurisprudence was ever designed to do the job the Court is now being asked to perform.