WeFootnote ¹ started the Kampala negotiations with a mixed set of preconditions. On the positive side, we had a definition of the crime of aggression, without brackets, adopted in February 2009.Footnote ² While the definition was always considered the easier part of the two big aspects of aggression – conditions for the exercise of jurisdiction being the other – this was no small achievement: most observers had originally been sceptical about the prospect of having any agreement in the Special Working Group on the Crime of Aggression. The support for the definition was solid at the time of its adoption, and it grew even stronger when it was subject to criticism and reservations in discussions later on.Footnote ³ There was a certain level of understanding for the concerns advanced, but the view prevailed that reopening the definition would inevitably lead to its total unravelling. States parties were, however, willing to address legitimate concerns outside the definition itself – an approach which led to the Understandings adopted in Kampala as part of the package on the crime of aggression. We could therefore go into the Review Conference with some confidence that we would be able to adopt the definition by consensus, and therefore used the formula ‘Definition Plus’ as the goal for an outcome.Footnote ⁴

Furthermore, the work leading to Kampala had been of high quality and created a collective belief that we were prepared for the discussions and therefore also for an outcome that in the end was bigger than had commonly been expected.

Nevertheless, there was also much scepticism and even anxiety. Irrespective of the level of preparedness of states to find an agreement on the crime of aggression, some were of the view that the International Criminal Court (ICC) itself was not yet ready to take on aggression cases.Footnote ⁵ Many states and to an almost greater extent non-governmental organizations (NGOs) also feared a divisive outcome or – for many the lesser evil – no outcome after a very divisive discussion. In some circles, it was communis opinio that a substantive outcome without a vote was not possible. And for many, a vote was per se – irrespective of its outcome and of the substance of a proposal on the table – anathema.

Before Kampala, two main big issues remained unresolved. The first concerned the question whether a proprio motu investigation and a state referral regarding a crime of aggression should be subject to a more restrictive jurisdictional regime than is the case for the other three core crimes. In essence, the question was whether some form of expressed consent by the alleged aggressor state should be required or not. The second issue concerned the role of the UN Security Council: should the ICC only be allowed to proceed where the Security Council actively determines that an act of aggression has taken place, or should there be alternatives so that the ICC could proceed even in the absence of such a determination? On both topics delegations held strong and seemingly irreconcilable positions, usually presented as positions of principle. The two issues were also to some extent interlinked, which further complicated the matter. The work done before Kampala had not led to an apparent narrowing of positions in this area.

As the basis of negotiations in Kampala we presented a Conference Room Paper that did not reduce the options on any of these issues, but offered a framework of a possible outcome package: enabling resolution, draft amendments, draft Elements of Crimes and draft understandings. A few relatively new ideas for elements of a final compromise were presented in a non-paper, including the idea of delaying the entry into force for the regime.

The Conference Room Paper was favourably received and discussed in a positive atmosphere, which allowed for some narrowing down in the first revision of the text. Three options for ‘jurisdictional filters’ in case of inaction by the Security Council were deleted: the ‘no filter’ option, but also the General Assembly and the International Court of Justice (ICJ) option. The discussion of the role of the Security Council was thus reduced to the core question: should the ICC's pre-trial judges be able to authorize the proceedings in the absence of a Security Council determination of aggression – or not? The second big issue, the question of aggressor state consent, which was closely linked to the modalities for entry into force of the amendments, was not affected by this revision. Several new ideas were floated in this regard, which required further discussion by states.

Most notably, Argentina, Brazil, and Switzerland (‘ABS’) proposed a new approach on the issue of entry into force, based on a separation of the triggers; the provisions on aggression investigations based on Security Council referrals would enter into force after only one ratification, following the procedure contained in Article 121(5) of the Rome Statute, while state referrals and proprio motu investigations would require ratifications by seven-eighths of states parties, in accordance with Article 121(4). Proprio motu investigations and state referrals would thus be deferred for a long time, if not indefinitely – but once activated, the Court would be allowed to investigate on this basis irrespective of any form of consent expressed by the alleged aggressor state, even with respect to a non-state party. This proposal met with mixed reactions. It created interest especially among those who favoured a protective regime that does not require aggressor state consent. It was at the same time also strongly criticized for not being consent-based, and for its legal technique.

The ABS initiative, which was complemented by proposals from Slovenia, proved crucial for the further course of the negotiations. On process, it illustrated that delegations were prepared both to assert ownership and to make serious and difficult compromises in order to reach an outcome. On substance, it introduced the idea of separate provisions for the triggers. The second revision of the Conference Room Paper reflected that separation, as it usefully focused the discussions on ‘aggressor state consent’ on the area where it belonged: state referrals and proprio motu proceedings.

Another important initiative was taken by Canada, which submitted a text that was strongly based on the principle of state consent, but complemented by an element of reciprocity and choice of filter. Both the ABS and the Canadian initiatives could not be ignored by the chair; the two camps espoused strong and, on substance, opposing policy views, but indicated willingness to make concessions to the other side.

To our surprise, a joint ABS/Canada proposal emerged on the day when the Working Group forwarded its report to plenary.Footnote ⁶ The proposal delayed the ICC's exercise of jurisdiction by at least five years after its entry into force for the first state party, and reflected a compromise approach to the issue of state consent: it excluded non-states parties altogether, and allowed states parties to opt out of the ICC's jurisdiction. The two parties to the conversation had thus, compared with their original positions, made very significant concessions. This compromise found considerable support among delegations, and the first non-paper of the presidentFootnote ⁷ reflected this progress, with significant modifications: it contained the opt-out procedure that was eventually adopted as part of the final package, although still in brackets. I presented it to the plenary as a sui generis solution on the basis of the acceptance already given by states parties under Article 12 of the Rome Statute,Footnote ⁸ which was possible due only to the specific placement of the crime of aggression in the Statute. On the question of filters, the paper kept the two main alternatives – that is, the Security Council filter and the Pre-Trial Chamber filter, with footnotes indicating possible variations.Footnote ⁹

This text also suggested, for the first time – but still in brackets – a simplification of the procedure for investigations based on a Security Council referral (now 15 ter), which would eliminate the previously implied need for the Prosecutor to go back to the Security Council for a determination of an act of aggression. The idea had emerged and to some extent been tested in bilateral consultations and seemed worth exploring in plenary. At this moment, we were also able to present an agreed text of the ‘Understandings’, facilitated by Germany. This was significant, because it wrapped up the consensus on the definition and enhanced the perception that a consensual agreement on a substantive package was possible.Footnote ¹⁰

Half a day later,Footnote ¹¹ I presented a further revised non-paper, with no more brackets, except with respect to the filters. It included in particular the ‘streamlined’ procedure for Security Council referrals (15 ter). On the central question of the filter for proprio motu investigations and state referrals (15 bis), the non-paper continued to reflect the two main approaches, with some modifications.Footnote ¹² Given the provisional agreement reflected in this text, the subsequent informal consultations focused strongly on the two filter approaches, and it was clear that a proposal of the president on this issue was required. At this point, less than 24 hours before the end of the Conference, my assessment of the political dynamic at the Conference was such that the submission of a text that provided for an alternative in case of Security Council inaction was the logical next step – although I was aware that this might raise serious objections. I therefore informally consulted the most important stakeholders in the late morning of Friday, 11 June, on the basis of a short non-paper. The paper suggested the Pre-Trial Division as an additional filter to the Security Council,Footnote ¹³ and sought to balance this choice in two ways: first, it made a specific reference to Article 16 of the Statute and the competence of the Security Council to suspend ICC proceedings. Second, it delayed the activation of the ICC's jurisdiction by at least seven years, and in doing so gave precedence to the Security Council filter: the ICC's jurisdiction under 15 ter would automatically be activated after seven years, unless states parties decided otherwise. The more controversial jurisdiction under 15 bis, on the other hand, would require an active decision by states parties. The feedback received during these consultations allowed for the conclusion that the choice of filter might find agreement, provided that we could find the right formula for activation. That in itself was a breakthrough of sorts, but the search for this formula proved more controversial than anticipated and consumed many hours of informal consultations in various formats.

The approach itself was introduced in plenary in a further revised and final non-paper on Friday afternoon, which had no brackets except for placeholders in 15 bis and 15 ter on delayed activation.Footnote ¹⁴ Direct talks between the most interested parties continued well into the evening with an extended and at times irrational argument over this aspect of the final package. In the end, they were unsuccessful. Given the high stakes and the late hour, I put a final proposal on the table, which today is the text of paragraph 3 of Articles 15 bis and 15 ter of the Rome Statute: the activation of both triggers is thus subject to a future decision of states parties, to be taken after 1 January 2017 by at least an absolute majority of two-thirds of states parties. I considered in Kampala, and still do today, that the issue of the formula for delayed activation – while important – was of significantly less relevance than some of the compromises that had been forged beforehand, in particular the opt-out regime for states parties, the wholesale exemption for non-states parties and the competence given to the Pre-Trial Division to authorize an investigation in the absence of a determination of aggression by the Security Council.

Few had expected – or even thought feasible – a consensus in Kampala on a comprehensive package on the crime of aggression. That this proved possible in the end was due to the very positive negotiating dynamic in Kampala, which found its most important expression in the willingness of all sides to make massive concessions. It was also recognized that this spirit of negotiations would probably be very difficult to re-create in the future.Footnote ¹⁵ And, indeed, the Kampala consensus reflects very significant compromises that all interested parties were prepared to make in order not only to finish the one task that states parties had left over from the Rome Conference in the best interest of the International Criminal Court, but also to make, collectively, a big step to advance international criminal law.