Hard Facts and the Threshold: What the Gambia v. Myanmar Should Not Decide

[His Excellency Filipo Tarakinikini is the Permanent Representative of the Republic of Fiji to the UN. He has spent more than two decades as an infantry officer and then another two in United Nations service. 

Geert-Jan Alexander Knoops is aprofessor of international law at the University of Amsterdam and defence counsel before the ICC, having spent more than three decades as defence counsel before the ICTY.

Gregory Rose is a professor of international law at the University of Wollongong.]

Note: Fiji did not intervene in The Gambia v. Myanmar. In this post we will address the doctrine rather than the outcome.

The Court concluded three weeks of merits hearings in The Gambia v. Myanmar on 29 January 2026 and is now in deliberation. It is the first full merits hearing in a genocide case in more than a decade, the first in which the Court will rule on a dispute brought by a non-injured State, and the first ever ICJ genocide case deciding culpability in circumstances where there has not been a separate prior judicial finding of guilt proved beyond a reasonable doubt by an international criminal tribunal. Whatever it decides will be, for a generation, the modern law of genocide.

Our concern is not with the outcome. It is with the reasoning, and specifically with how far that reasoning will be asked to travel.

The Lesson of Gotovina

In April 2011, the ICTY Trial Chamber convicted Ante Gotovina after employing what became known as the “200 metre standard”. The Chamber relied on a rule it had itself established: that artillery impacts falling more than 200 metres from a legitimate military target evidenced unlawful indiscriminate shelling. The standard appeared nowhere in the evidence, in the pleadings, or in the law of armed conflict. It was constructed because the Chamber needed something to bear the weight of the inference it wished to draw. Nineteen months later the Appeals Chamber found the standard to be without foundation, and Gotovina was acquitted.

The episode is worth recalling because it is the clearest modern illustration of a recurring hazard. Where the facts before a tribunal are reflecting the overwhelming complexity of urban warfare and the applicable evidentiary standard is demanding, there is a structural temptation to relieve the pressure by adjusting the standard rather than by confronting what the evidence will and will not carry. The adjustment rarely survives appellate scrutiny. But the ICJ has no appellate chamber. What it decides in Myanmar will not be legally corrected; it will be cited.

Four Questions, and How Far the Answers Will Travel

The questions put from the Bench during the January hearings indicate that the Court is alive to precisely the issues on which its reasoning will be most portable.

First, Judge Hmoud asked whether individual criminal responsibility and State responsibility for genocide constitute distinct regimes, such that the dolus specialis required for attribution to a State differs from that required for attribution to a person. This is the most consequential question before the Court. The Convention describes genocide, in its title, its preamble and Article I, as a crime. 

If State responsibility for that crime can be established on a standard materially lower than the criminal standard, international criminal jurisdiction over genocide becomes superfluous. Adopting a lower evidentiary standard for State responsibility would create a strong incentive for strategic forum shopping. Applicant States could seek a finding of genocide before the ICJ rather than pursue the far more demanding criminal process before the ICC. The potential rationale for a differentiation between the evidentiary standard for a state responsibility and the standard for individual criminal responsibility, being that the latter is about the establishment of individual criminal guilt, seems not convincing.  An ICJ determination that a State committed or failed to prevent genocide would carry significant political and legal weight and open the door to subsequent criminal proceedings against individual leaders. The rationale for a differentiation between the evidentiary standard for state responsibility and for individual criminal responsibility, would then become less relevant, simply because state responsibility, findings are likely to be relied upon by prosecutors and courts as part of the broader evidentiary context for individual guilt. For that reason alone, there is no rationale for applying different standards of proof to the same allegation of genocide. 

Second, Judge Cleveland asked what weight should attach to NGO reports tendered in evidence. 

Third, Judge Gomes Robledo asked whether adverse inferences may be drawn from a respondent’s refusal to admit international investigative bodies. Both questions concern the architecture of proof, and both will be answered once, in Myanmar, for every genocide case that follows.

The evidentiary question is not a technicality dressed up as principle. If reports of United Nations fact-finding missions and commissions of inquiry acquire presumptive probative status by virtue of their institutional provenance, then a body of material assembled to a standard expressly lower than any court’s — reliant on secondary sourcing, frequently unable to verify, produced under mandates that are themselves the outcome of political decision — enters the record with a weight it was never designed to carry. Bassiouni identified this problem a quarter of a century ago; it has not been solved since. The answer is not to exclude such material but to require that it be weighed individually and critically on its methodology rather than its letterhead. The same discipline should apply to reports produced by non-governmental organisations, whose missions, funding and prior advocacy positions vary enormously and are relevant to weight.

The adverse-inference question carries a similar risk of over-extension. A State may decline access to an international body for reasons that have nothing to do with consciousness of guilt: for example, apprehension of methodological unreliability, of politicised mandates, of predetermined conclusions. Where the ultimate finding must rest on an inference that genocidal intent is the only reasonable inference available ignoring alternative explanations cannot bear that weight.

Fourth, and underlying all of this, is Article II(c) of the Genocide Convention. “Deliberately inflicting on the group conditions of life calculated to bring about its physical destruction” is the most elastic of the enumerated acts, and it is where an expansive reading would do the most work. The words “deliberately” and “calculated” are not ornamental, rather they import the specific intent requirement into the actus reus itself. A reading that permits tragic conditions of life to infer their own intent collapses the distinction the drafters were at pains to construct.

The Distinction that Must be Preserved

Here is the point we most want to make, and we lodge it deliberately before the judgment rather than after.

The Gambia v. Myanmar concerns clearance operations directed at a denationalised, unarmed minority. There was no organised armed adversary embedded within the Rohingya population. There were no dual-use objects in contention, no combatants operating from civilian infrastructure, no question of proportionality in urban targeting. Whatever the Court holds about inferring genocidal intent on those facts is a holding about those facts.

It cannot be transposed, without more, to intense urban warfare against a structured armed force that operates from within civilian infrastructure. The two situations differ not only in degree but in kind, because in urban warfare there are more likely to exist alternative explanations for civilian death that do not exist in the first. The only reasonable inference test — laid down in Bosnia in 2007 and reaffirmed in Croatia in 2015 — does its essential work precisely at that point. It asks whether any explanation other than intent to destroy the group can reasonably account for a pattern of conduct. Where a State is prosecuting an armed conflict against an embedded adversary, and has taken the precautions international humanitarian law requires, alternative explanations exist by construction. That is not a loophole. It is the test operating as designed.

We say this now because a distinction drawn before a judgment is a proposition of law, and the same distinction is an excuse.

What we are Not Arguing

We are not arguing that Myanmar should prevail, and we do not suggest that the Rohingya facts fail to meet the existing threshold. They may well satisfy it. Our submission concerns the route, not the destination. A finding of genocide reached by applying the settled dolus specialis standard rigorously to compelling evidence is the Convention working properly. A finding of genocide reached by relaxing the standard because the evidence is compelling, amounts to subverting the Convention. 

Equally debatable is the argument of some of the interventions before the Court in the Gaza proceedings. For instance, Ireland has invited the Court to find genocidal intent where a reasonable person would have foreseen destruction of the group as a likely consequence of a State’s conduct. Whatever the merits of that proposition as policy, it amounts to a negligence standard. Foreseeability is not intent; it is the absence of intent, supplied by construction. To adopt it would be to abolish dolus specialis while retaining the word that depends on it.

The argument to adopt a “holistic” approach to interpretation of the Convention is subject to the same objection in a more general form. Articles 31 and 32 of the Vienna Convention state the customary rules, as this Court has confirmed in this very context. There is no principled basis for a bespoke interpretive method reserved for this treaty in this circumstance alone, and the effect of adopting one would be to leave the substantive content of the most serious prohibition in international law contingent on a novelty. Where a legal standard is treated as sufficiently flexible to accommodate any conclusion, adjudication ceases to constrain outcomes and begins merely to ratify them. Courts that reach that condition do not increase their authority, they exhaust it by subverting the rule of law.

Why this Matters to States that Supply Soldiers

There is a constituency in this debate that has been almost entirely silent, and it has the most to lose.

Fiji has contributed troops to United Nations operations continuously since 1978 — in Lebanon, Sinai, Iraq, Timor-Leste, Golan, and elsewhere — at a scale wholly disproportionate to its population. Professional armed forces generate After-Action Reviews as a matter of standing practice: contemporaneous records produced for command accountability and legal compliance, not for litigation. If such records are treated as inherently self-serving while UN institutionally produced reports are treated as presumptively more reliable, the incentive to generate After-Action Reviews honestly, or at all, degrades. This is a legal danger to every professional military that takes compliance seriously, and it falls hardest on those that document most carefully.

The second consequence is more direct. If a State that volunteers forces for an international operation — to disarm an armed group, to stabilise a collapsed authority — thereby exposes itself to genocide proceedings brought by that group’s patrons, small troop-contributing States will conclude that the risk is not worth carrying. They will not announce this. They will simply stop offering. The Convention was not drafted to produce that result, and a Court that produces it will have weakened the prevention of atrocity in the name of its punishment.

Conclusion

The reason that five genocide cases have been brought before the Court in the past seven years, but only two cases in the first 50 years of the Genocide Convention, might be a signal that the genocide jurisdiction can be misused to get around the fact that the Court has no jurisdiction over armed conflict.

The threshold for genocide is high because the word means something specific and serious. Every dilution offered is paid for by the next case, and the one after that, in which the word is asked to do work it can no longer do.

Genocide law is prescribed and the Court should apply the standard it has. It should not adjust it because the facts are terrible. That the facts are terrible is precisely when the standard is worth having.