Four Mechanisms, One Test
Bosnia, Kosovo, Bangladesh and Syria, and the question of what makes an atrocity trial credible
The design debate is the wrong debate
On 16 September 2026, the Kosovo Specialist Chambers, sitting in the Hague, will deliver its trial judgment in the case against Hashim Thaçi, Kadri Veseli, Rexhep Selimi and Jakup Krasniqi, after repeated postponement. Five weeks before that, on 11 August 2026, the Fourth Criminal Court in Damascus sentenced Bashar al-Assad to death after nine hearings. Ten months earlier, on 17 November 2025, the International Crimes Tribunal in Dhaka sentenced Sheikh Hasina to death in her absence. Three mechanisms, three quite different institutional designs, and a single question that none of them can avoid.
Much of the debate about atrocity crimes still turns on architecture. Should accountability be pursued domestically, through a hybrid court, or before an international tribunal? The question has occupied a generation of practitioners and has produced a great deal of valuable comparative work. On the evidence of the last two years, it is also, largely, the wrong question. Whatever the label on the door, the same problem emerges. A trial must be fair, and it must be seen to be fair, or it will be neither legitimate or credible. It must be transparent and publicly accessible. It must address public demands for justice, but not be influenced by public demands for swift, meaningless, convictions. Legitimacy is not conferred by a statute or by the nationality of the judges. It is earned in the transparent and fair conduct of the proceedings and in the quality of the reasoning that follows.
Four examples make the point, and they do not divide along the lines one might expect.
A Note of Transparency – I served in some capacity in all four
At the beginning of my professional career, in Bosnia, I served first in the Defence Office and then later in the Prosecution. Subsequently, in Bangladesh, I again served first in the defence and was later appointed to the Prosecution. In Kosovo, I served exclusively on the defence, and have remained a critical advocate of the process of lacking any domestic capacity and being an unaccountable institution to the people of Kosovo. Finally, in Syria, I have worked extensively with victims since the revolution started and have worked on training the exiled judges and prosecutors whose task is now to bring accountability for the crimes of the Assad regime.
Bosnia: the hybrid that broadly worked
I am starting with Bosnia, as I still consider it the most successful accountability mechanism, certainly of all four cited in this piece. I recognise that having been involved in its inception and development, I am somewhat biased, but it is a view broadly shared.
The War Crimes Chamber of the State Court of Bosnia and Herzegovina became operational in March 2005. It was built to take referrals from the International Criminal Tribunal for the former Yugoslavia under Rule 11bis and to absorb the far larger body of cases that the ad hoc Tribunal in the Hague was never going to reach. It sat in Sarajevo. It applied the newly adopted Criminal Code of Bosnia and Herzegovina. Its panels were mixed, with international judges sitting alongside national judges, and a Special Department for War Crimes was created within the Prosecutor's Office on the same model. Critically, the international element was designed from the outset to be temporary. It was phased out, rather too aggressively in my view, but by the end of 2012 the Chamber was a wholly national institution, a decision based more on funding than capacity.
It is easy to be unsentimental about what followed. Fewer than 200 final judgments in twenty years is not a triumph, particularly when one considers what was achieved in the first decade. Human Rights Watch recorded 222 war crimes cases pending before all courts in Bosnia as of September 2025, with only eight first instance judgments delivered in the first eight months of that year. The deadlines under the Revised National War Crimes Strategy have slipped repeatedly. Victims of conflict related sexual violence are still being made to pursue compensation through the civil courts, and in Republika Srpska they are being charged costs for the privilege. Defendants and witnesses are dying of old age. Three decades after Srebrenica, accountability in Bosnia remains slow and incomplete, and the reception given to Ratko Mladić’s death this month shows how little the process has shifted the politics of denial.
And yet nobody in Bosnia argues that the Court of Bosnia and Herzegovina is a foreign imposition, that is apart from the hardline government in Republika Srpska and those that continue the politics of denial and Serbian victimisation. It is a Bosnian court, applying Bosnian law, staffed by Bosnian judges and prosecutors, sitting in the country where the crimes occurred, and its judgments are available to the communities they concern. Its failures are failures of resource, capacity and political will. They are Bosnia's failures, and they are in principle Bosnia's to fix. That is not a small thing. The transition from hybrid to domestic was messy and underfunded, but it happened, and it left something behind.
Kosovo: procedure without ownership
The Kosovo Specialist Chambers are the mirror image. They were created by the Kosovo Assembly in August 2015, amid pressure from the EU and US, but they have a statute and rules of procedure, and sparingly apply Kosovo law, and they are not formally part of the Kosovo judicial system. On paper they are the most domestic of the hybrids. In every other respect they are not. They sit in The Hague, in another country, far from the communities the crimes concerned. Every judge is international. Every prosecutor is international. There are no Kosovo nationals on the bench, in the Registry’s senior positions, or in the Specialist Prosecutor’s Office at any level that matters. In fact, they do not employ a single Kosovo national in any post. That will remain a stain on its legacy.
I appeared before the Chambers on behalf of the defence, and nothing that follows is a criticism of the professionalism of individual judges, individual prosecutors, or of the Registry, which is considerable. The procedural protections available to an accused before the Specialist Chambers are, on paper, among the most generous of any atrocity court in operation. Disclosure regimes, defence resourcing, the right to be present, the presumption of innocence, appellate review: the architecture is sound, in theory. As with everything, it is the practice matters that matters, not the theoretical framework or the illusory protection of fundamental rights.
That is precisely the point. Procedural rigour has not produced legitimacy. In Kosovo the Chambers are widely regarded as an instrument imposed to balance the moral ledger of the war rather than to try the crimes of it, and that perception is not confined to the political class. Most hearings are closed or subject to limited reporting. Tanscripts appear weeks late and heavily redacted. Protective measures have rendered victims invisible, so that the public never hears the accounts on which the prosecution rests. The Thaçi trial has run since 2023, the verdict date has slipped more than once, and each postponement has been received in Prišhtinë as evidence of a court that is answerable to nobody.
A court can be scrupulously fair and still fail, if the society whose crimes it is trying has been designed out of the process. Transparency is not a courtesy extended to the public. It is a component of the fairness itself.
Bangladesh: the demand for prompt justice
Bangladesh is the domestic model under maximum political pressure. The International Crimes Tribunal was first conceived in 1973 and there was pressure to internationalise it, but ultimately it was shelved before it begun in the interest of peace and stability. This was, in my view, patently wrong. It deprived victims of justice and ultimately served as a catalyst for internal conflict over the next four decades. It revived was revived by Sheikh Hasina in 2010 and a number of defendants were put on trial, and many summarily executed, in circumstances that can only be described as a flagrant denial of justice and a breach of both national and international law.
Following the uprising of August 2024, an interim government was put in place and the prosecution brought charges to try members of the government, including Sheikh Hasina, that the uprising had displaced. On 17 November 2025, it convicted Sheikh Hasina of crimes against humanity and sentenced her to death, together with the former Home Minister, in proceedings conducted in their absence. In February 2026, the Awami League was barred from contesting the general election, which the Bangladesh Nationalist Party won with a two thirds majority. India has consistently declined to surrender Hasina, although she has vowed to return to Bangladesh in December.
I served as Special Adviser to the Chief Prosecutor of that Tribunal, and I stepped down from that role in December 2025. I have been critical about the Tribunal’s defects since 2011, when the accused were largely members of the political opposition and the government was Hasina’s, and I see no reason to be less consistent now that the political valence has reversed. The 1973 Act as originally enacted made no provision for trial in absentia. Section 10A was inserted by amendment in 2012, and the Rules require no more than notice by newspaper advertisement before a trial proceeds without the accused. There is no requirement of an express and unequivocal waiver. There is no guaranteed right to a retrial on apprehension. The maximum sentence is death. I have consistently opposed the death penalty, in all circumstances, and I do not support trials in absentia, particularly when fundamental rights are curtailed. It remains my view that the evidence against Sheikh Hasina was overwhelming and she should have been tried in her presence. A trial in absentia offers little to the victims and much less to a historical record. If she is to return in December then she should be re-tried in her presence with the assistance of counsel.
The pressure that produced the in absentia trial is entirely understandable. Around 1,400 people were killed in the crackdown of July and August 2024. A population that has just removed a government by force does not want to be told that accountability is a matter of years. The demand for prompt justice is real, and it is legitimate as a demand. It is not, however, a standard of proof, and a tribunal that treats it as one has already conceded the argument to those who say the outcome was arranged in advance. The IBAHRI and the UN human rights mechanisms have both said as much. The real test will be what happens now. This is a great opportunity for the current government to ensure that truth, justice and accountability are cornerstones of its policy.
Syria: nine hearings
Syria has now done the same thing faster. On 11 August 2026 the Fourth Criminal Court in Damascus sentenced Bashar al-Assad, his brother Maher and their cousin Atef Najib, along with six others, on charges including murder, torture, unlawful detention, war crimes and crimes against humanity arising out of the repression at Daraa in 2011 and what followed. The Assad brothers were tried in absentia from Moscow. The trial ran to nine sessions between 26 April and 11 August. There is no evidence that the absent accused received formal notice or that counsel was assigned to them. The 396-page judgment has not been published.
There are further difficulties beneath the procedure. The Syrian Penal Code does not codify crimes against humanity or war crimes, and the content of command responsibility in Syrian law is unclear. A court that convicts on those charges without publishing its reasoning has given nobody the means to assess how it did so. Meanwhile the National Commission for Transitional Justice established by decree in May 2025 is confined to violations committed by the former government, which tells Syrians harmed by anyone else that their cases fall outside the frame.
A death sentence against a man under Russian protection changes nothing on the ground. What it does establish is a domestic precedent that a Syrian court may convict of crimes against humanity in nine sessions on an unpublished judgment. That precedent will not stay attached to the most notorious name in the country. It will be applied to defendants nobody has heard of.
Trials in absentia, and what a death sentence forecloses
Trial in absentia is not prohibited outright. Article 14(3)(d) ICCPR guarantees the right to be present, and General Comment No. 32 accepts that proceedings in absentia may be permissible in exceptional circumstances. The conditions, though, are cumulative and they are strict: the accused must have been notified in terms that leave no doubt, all reasonable steps must have been taken to inform him, counsel must be guaranteed, and there must be a right to a fresh trial once he is in custody. The Strasbourg line from Colozza through Poitrimol, Krombach and Sejdovic says the same. Silence is not waiver. The ICTY, the ICTR and the Special Court for Sierra Leone declined to try anyone in absentia at all. The Special Tribunal for Lebanon was the exception, and it carried an express statutory right to retrial.
It is questionable whether Dhaka or Damascus meets those conditions, and the reason is structural rather than accidental. A state resorts to trial in absentia precisely because it knows the accused will never be produced. The right to a retrial on apprehension is therefore theoretical, and everybody involved understands that it is theoretical. What is left is a sentence intended to be symbolic. When the symbol is death, the objection is not merely that the process is unfair. It is that the sentence forecloses the trial that might otherwise one day have taken place. No state that has abolished capital punishment may lawfully surrender a person to face it, so the sentence hardens the position of every jurisdiction in which the accused might be found and hands him a grievance narrative he could not have manufactured for himself. Bangladesh has made Hasina harder to extradite. Syria has made Assad harder to reach. In both cases the process has purchased a headline at the cost of the outcome.
What actually matters
Three propositions follow, and none of them is about institutional design.
First, national ownership is necessary but not sufficient. Bangladesh and Syria are wholly domestic processes, entirely owned, and they are the least credible of the four. Ownership without procedural discipline produces victors' justice with a court's letterhead.
Second, procedural rigour is necessary but not sufficient. The Kosovo Specialist Chambers have the discipline and lack the ownership, and the consequence is a court whose judgments are unlikely to be accepted in the society they concern whatever they say on 16 September. A judgment that nobody in Kosovo believes is not a contribution to accountability. It is a contribution to the grievance that follows it.
Third, the Bosnian compromise, mixed panels sitting in the country, applying domestic law, with an international component designed to withdraw, remains the least bad model available. Its record is not what it promised, but its failures are of resource and political will rather than of conception, and they are addressable. That cannot be said of a court sitting in another country with no national participation, and it cannot be said of a tribunal that sentences absent defendants to death on newspaper notice.
The test is the same in every case, and it is not a demanding one to state. Was the accused genuinely able to answer the case? Can the reasoning be read? Can the affected community see the process and recognise it as their own? A mechanism that satisfies those three conditions will be criticised, appealed and resented, as courts always are, and it will still be believed. A mechanism that satisfies none of them will not become credible because it was quick, and it will not become credible because the defendant is guilty.
A court that will not publish its judgment has already told you which kind it is.
Comments