The Law of ‘Never Again’: Transitional Justice and the Transformation of the Norm of Non-Recurrence

This article analyses the transformation of guarantees of non-recurrence (GNRs), the least developed pillar of transitional justice (TJ), and sets a legal and conceptual foundation of the norm for TJ theory and practice. It draws out key characteristics of GNRs including the norm’s various contents and contexts, stressing its exceptional fu-ture-oriented nature in international law. The article investigates conceptual origins of preventing non-recurrence in the early developments of TJ and the recent normative expansions undertaken by the UN Special Rapporteur. The main contributions of the article are establishing GNRs as normatively distinct in TJ and identifying transfers of local-level advocacy from Latin America to general norm creation. Finally, the article proposes a tension between decontextualizing the norm content to make it universally applicable and recent attempts to normatively expand the norm and improve its con-text-speciﬁcity, and discusses its potential consequences for TJ practice.


INTRODUCTION
Ensuring non-recurrence of violations of international law is a norm with a most curious journey. First used to solve diplomatic disputes, in more recent years guarantees of non-recurrence/non-repetition (GNRs) were codified as aspects of state responsibility for internationally wrongful acts, a form of reparations for international human rights law (IHRL) and humanitarian law violations and a right of states as well as individual actors. In 2011, when the office of the UN Special Rapporteur for the promotion of truth, justice, reparation and guarantees of non-recurrence (hereafter: Special Rapporteur) was created, the four pillars of the field of transitional justice (TJ) were confirmed. Subsequently, the central goal of preventing future repetitions of violence in TJ, also known as the 'Never Again' promise, is now phrased as GNRs and separated from reparations, contrary to IHRL.
In this article, I respond to the identified gaps about GNRs in TJ scholarship and aim to provide a conceptual foundation for the norm of non-recurrence in TJ. I trace the development of GNRs from the early examples of inter-state disputes and public international law (PIL) to non-recurrence as victims' rights in IHRL and finally, early transitional justice. Throughout these sections, I stress the competing understandings of GNRs across legal regimes and in TJ and observe that, although states have been reluctant to seek specific GNRs before the International Court of Justice (ICJ), UN agencies and most notably transitional justice scholars and actors have taken up the task of affirming GNRs as a norm with differentiating non-exceptional and contentspecific characteristics. These multiple actors have been very progressive in developing GNRs, demonstrating a shift towards an expanded scope of both the right and obligation to guarantee non-repetition of violations. Most importantly, I argue that it is TJ that inspired the content/contextual expansion of GNRs in international human rights law and not vice versa, therein substantially transforming the previously ambiguous and exceptional norm into a set of measures of legal and institutional reform.
Several key questions about GNRs in transitional justice are answered in this article. First, what is, as Pablo de Greiff phrases it, 'the offer': 12 what content does a guarantee of non-recurrence entail across legal regimes and in TJ? Second, how did GNRs secure their position as one of the pillars of TJ? Third, what is the relevance of GNRs' legal origins and the developed conceptual mismatches and overlaps for TJ practice? The article makes three core contributions to TJ literature. Firstly, it does so by discovering the influence of the transitional Latin American experience, most notably Chile, on the transformation of the norm of non-recurrence in IHRL. Secondly, it accounts for GNRs as becoming normatively different in transitional justice than in PIL and IHRL and explores the relevance of this distinction in international TJ advocacy and practice. Lastly, it opens a conversation about norm revolution, or rather confusion, of GNRs in and by transitional justice, and the potential effects on future practices. The article starts with a review of GNRs in PIL, followed by an overview of the transformation of GNRs in IHRL. These sections remain rather brief and only capture the main developments regarding GNRs in the two legal regimes. The article then shifts its focus back to transitional justice, where I examine the conceptual origins of equating non-recurrence to measures of institutional reform, and further trace the transformations of the norm undertaken by the Special Rapporteur. Finally, I discuss the consequences of decontextualized norm transformation and attempts to recontextualize it in TJ practice and call for further empirical studies of GNRs.

STATE RESPONSIBILITY TO ENSURE NON-REPETITION
The recorded story of GNRs began nearly three centuries before that of transitional justice. Since then, GNRs have developed as a norm of exceptional character in PIL. Numerous examples of past diplomatic practice point to a wide acceptance of injured states' right to demand these guarantees, suggesting, by way of a customary norm, that assurances and GNRs could be a legal consequence of an internationally wrongful act. 13 One of the earliest documented practices takes us back to 1707, when the Russian Ambassador to the newly created United Kingdom, Andrey Matveyev, was arrested by his creditors, mishandled and placed in prison. This was done according to the English law that at the time did not offer protection to aliens from imprisonment for debt. 14 After Matveyev's angry departure from England, an Act of Parliament called the Diplomatic Privileges Act was adopted in 1708 to prevent the recurrence of such incidents in which foreign ambassadors could be physically injured. 15 The Diplomatic Privileges Act is a less common example of the injuring state practising its obligation of conduct through the adoption of specific measures or actions. Besides new legislation, other such specific measures of non-recurrence taken to solve inter-state disputes included apologies and instructing the injuring state's agents to act in specific, improved ways. 16 Most commonly, however, states demanded assurances -as opposed to guarantees -of non-repetition: that is, expression of mere moral commitment to non-recurrence. By way of example, in the 1904 Dogger Bank incident, the UK requested that Russia provided 'security against the recurrence of such intolerable incidents.' 17 The legal obligation was that of result, and the sole absence of repetition of the same event against the same state would suffice. This line of state practice thereby shows that although the idea of legislative reforms as a possible guarantee of non-recurrence existed and subsequently persisted, the specific content of GNRs was predominantly absent.
Two main sources that codify GNR in PIL are the ICJ's decision in the LaGrand case 18 and the International Law Commission's (ILC) Articles on Responsibility of States for Internationally Wrongful Acts 19 (ARSIWA). Both were issued as recently as 2001. The LaGrand case concerned alleged breaches of Article 36 of the Vienna Convention on Consular Relations, which the US committed by failing to inform two foreign detainees, the German nationals and brothers LaGrand, of their rights before sentencing them to death and executing them. Germany requested a guarantee that the opposing party would not repeat these violations in the future and would ensure that both its domestic law and practices allowed for an effective exercise of the rights included in the Convention. Germany's request signals that it would not have been satisfied with the more common verbal assurance of non-repetition or a symbolic apology. Instead, Germany concluded that more was needed and that more meant specific GNRs, concrete steps that would prevent the automatic reproduction of violations of the Vienna Convention. In addition to recognizing general GNRs such as apologies, the ICJ acknowledged the existence of specific guarantees that would trigger amendments in laws or policies; 20 nevertheless, it did not go further in explaining what particular steps the US would need to take to comply with such an order, leaving the whole concept somewhat ambiguous. 21 Therefore, the choice of the form that GNRs take was left to the injuring state. 22 Observing the more recent cases together with LaGrand, 23 there seems to be an agreement at the ICJ that assurances and guarantees of non-repetition are a possible, but not a necessary consequence of committing an internationally wrongful act for the injuring state, and that the injured state might be entitled to seek both general and specific guarantees that the breach will not be repeated. The LaGrand case also implies that GNRs are qualitatively distinct from other traditional types of reparations which states seek to restore the status quo; GNRs, on the contrary, are 'forward-looking,' and invested in the prevention of future violations by changing -and not restoring -the status quo. 24 In the drafting history of the ARSIWA, the idea of state responsibility to prevent the repetition of injurious acts was present from the early stages. Already in 1961, F.V. Garcia-Amador, an independent expert hired by the ILC, included measures to prevent repetition in Article 27 of his draft Articles on state responsibility for injuries caused to aliens. 25 All subsequent independent experts followed Garcia-Amador's ideas in their recommendations about non-repetition. Still, James Crawford, who was one of the drafters, admitted that GNRs were one of the most debated provisions before the adoption of ARSIWA. 26  In the final version of the ARSIWA of 2001, a state responsible for an internationally wrongful act has an obligation to not only cease the act but, under Article 30(b), to 'offer appropriate assurances and guarantees of non-repetition, if circumstances so require.' The very last clause of that Article, previously changed from 'when appropriate,' indicates, as the minutes of the Drafting Committee show, that GNRs are not a necessary legal consequence of any breach of international law. 29 Hence, the provision on GNRs was drafted with flexibility in order to, according to the commentary, 'prevent the kinds of abusive and excessive claims' of GNRs certain states made 'in the past.' 30 Whether assurances and GNRs will be awarded depends on the circumstances of the case, including the nature of the breach and the obligation. Despite previous classifications of GNRs as a form of satisfaction or a separate type of reparation, the final version of the ARSIWA defines them as neither. Namely, Article 34 delineates different forms of reparation as restitution, compensation and satisfaction -to be applied individually or in combination -but does not include GNRs under satisfaction. In the commentary, however, the ILC stated it did not make an exhaustive list of measures of satisfaction, and that GNRs could take the form of satisfaction, giving the example of repealing the legislation which brought about the violation in question as both a GNR and measure of satisfaction. 31 The overarching explanation for such a decision was that international practice has not been univocal when it comes to the content of GNRs. Nevertheless, placing GNRs within Article 30 implies norm autonomy due to their future-looking focus on prevention.
What both the ICJ's case law and the ARSIWA establish is that GNRs are exceptional; they are not a necessary consequence of a breach of international law and are flexible in content. The fact that neither the ARSIWA nor the ICJ's jurisprudence so far tells us much about specific guarantees can be interpreted as purposeful and in line with the non-uniform state practice. Most importantly for the future developments of GNRs in transitional justice, in PIL they are not only exceptional in their triggering of state obligation but also their nature; they are construed as qualitatively different, 'forward-looking,' progressive and preventive.

GUARANTEES OF NON-RECURRENCE AS A VICTIM'S RIGHT
Thus far, this article establishes the relevance and exceptional application of GNRs for states, should they breach an international obligation they have towards another state or the international community. However, guaranteeing non-recurrence has, as a legal concept, gained much more prominence in international human rights law, a regime geared towards individuals, which has expanded the content, context and state obligation to ensure non-recurrence. Experts in this regime drew from the ARSIWA to fortify their claims about state responsibility to ensure non-repetition of violations, despite the distinct nature of such violations. In the UN Basic Principles and Guidelines on the Right to a Remedy and Reparation for Victims of Gross

Violations of International Human Rights Law and Serious Violations of
International Humanitarian Law (hereinafter the Basic Principles), 32 GNRs have been conceptualized as a right of the victim. 33 The position GNRs occupy in this document largely stems from the work done by Theo van Boven and M. Cherif Bassiouni, two experts who were mandated by the UN to conduct studies on the right to remedy. Van Boven's first comprehensive set of principles was issued in 1993 and went through minor changes in the subsequent years, eventually differentiating between five forms of reparation: cessation, restitution, compensation, satisfaction and GNRs. 34 How did GNRs, previously treated as a right of an injured state and distinct from reparations, end up there?
That victims of human rights violations were entitled to an effective remedy was already cemented in several human rights documents predating van Boven's principles, such as Article 8 of the Universal Declaration of Human Rights, as well as Article 2(3)(a) of the International Covenant on Civil and Political Rights. Additionally, regional human rights conventions have their own provisions protecting this right, 35 and van Boven used all of these sources to fortify his drafts. Since none of these documents mention GNRs in any form, a likely explanation for their inclusion lies in van Boven's references to the work of the ILC on the ARSIWA which was, at that point, still in progress. Admitting that the ILC's draft articles are suited for inter-state disputes, he nevertheless found particularly useful the ILC's identification of different types of reparations. 36 For van Boven, however, GNRs and measures of satisfaction are one and the same, both enshrined in Article 11.
Most notably, van Boven drafted several specific measures that could be applied to meet the goal of preventing the recurrence of violations. This was the first time GNRs were specified as concrete measures in a UN document and exactly what the ILC shied away from doing. Bassiouni's 2000 draft articles on satisfaction and guarantees of non-repetition, faithful to van Boven's definition, expanded previous guidelines by adding two entirely new principles of GNRs without citing any sources of these ideas in the commentary. In this new draft, it was acknowledged that the principles incorporate some emerging norms and standards and aspire to anticipate future developments in international law. 37  The Law of 'Never Again' 7 was made to reflect existing international obligations ('states shall') in contrast to emerging norms ('states should'). 38 GNRs fall under the latter category, which suggests they were not a fully conceptualized international obligation at the time in this legal regime. During this period, GNRs also made appearances in other IHRL developments, emerging as a set of measures to reform state laws and institutions. Notably, when Louis Joinet drafted the principles on combatting impunity in 1997, he recommended a set of measures preventing victims from enduring new violations. Joinet included three broad categories of legal and institutional reform for non-recurrence, which were separate from the right to reparation: the disbandment of paramilitary groups; the repeal of emergency legislation and recognition of the inviolability and non-derogability of habeas corpus; and removal from office of senior state officials implicated in serious human rights violations. 39 Diane Orentlicher was later hired to update these principles 'to reflect recent developments in international law' which resulted in a report published in February 2005. 40 In her report, Orentlicher has regrouped reparation and GNRs together since 'human rights treaty bodies have often treated GNRs as a component of reparations,' including the then-most recent draft on the Basic Principles. 41 Institutional reform included under general principles of GNRs, according to Orentlicher's draft, should advance the objectives of consistent adherence by public institutions to the rule of law, repeal of legislation that contributes or authorizes violations of human rights and/or humanitarian law, civil control of military and security forces and intelligence services and disbandment of parastatal armed forces and, finally and distinctly, reintegration into society of children involved in armed conflict. 42 Several months after these impunity principles were updated, the Basic Principles were finalized. In this revised set of principles, GNRs, surprisingly and, I note, contrary to van Boven's and Bassiouni's drafts, appeared as a self-standing measure of effective reparation, along with restitution, compensation, rehabilitation and satisfaction, confirming Orentlicher's classification. This final conceptualization runs contrary to the ARSIWA where GNRs are deliberately separate from reparations as they 'look to the protection or maintenance of the legal relationship' and are, in that sense, presupposing a risk of recurrence. 43 Considering the importance of the Basic Principles in later TJ practices, it is worthwhile listing all the included measures of GNRs. Specifically, under Article 23 of the Basic Principles, GNRs should include, 'where applicable, any or all of the following measures': 38  1. Ensuring effective civilian control of military and security forces; 2. Ensuring that all civilian and military proceedings abide by international standards of due process, fairness and impartiality; 3. Strengthening the independence of the judiciary; 4. Protecting persons in the legal, medical and healthcare professions, the media and other related professions and human rights defenders; 5. Providing, on a priority and continued basis, human rights and international humanitarian law education to all sectors of society and training for law enforcement officials as well as military and security forces; 6. Promoting the observance of codes of conduct and ethical norms, in particular international standards, by public servants, including law enforcement, correctional, media, medical, psychological, social service and military personnel, as well as by economic enterprises; 7. Promoting mechanisms for preventing and monitoring social conflicts and their resolution; 8. Reviewing and reforming laws contributing to or allowing gross violations of international human rights law and serious violations of international humanitarian law. 44 Although in their content the final Basic Principles contained only minor departures from Bassiouni's 2000 guidelines, their novel positioning as a standalone form of reparations points to further norm autonomy. The listed measures, as demonstrated, largely focus on laws, institutions and state officials. To ensure non-repetition of human rights violations, therefore, by and large means to reform the state and its institutions. Bassiouni's commentary from 2006 suggests that for each of these measures the drafters took into consideration previous decisions and recommendations by the UN Human Rights Committee and the Inter-American Court of Human Rights (IACtHR), in addition to UN General Assembly Resolutions. 45 The Basic Principles are a crucial development in the story of GNRs in TJ. The principles provide this previously purposefully ambiguous and flexible norm with a number of specific measures that a state found in breach of human rights and humanitarian law might have an obligation to provide. This document challenges and expands not only state obligation but also the contents and contexts of GNRs. Reparations are usually meant to make amends to the victim(s); however, including GNRs suggests that they, through legal and institutional reforms, are meant to benefit the entire affected community. As far as contexts are concerned, GNRs previously applied in cases of typically individual breaches in interstate relations but have now become operational in the context of gross violations of international human rights (and humanitarian) law. The nature and gravity of such violations imply background circumstances of war, civil conflict or state oppression. This implication has been reflected in practice. By way of example, in the Ugandan peace agreement between the Government and the Lord's Resistance Army, GNRs are referred to as a measure of reparations in accordance with the Basic Principles. 46 The principles, therefore, make GNRs well-suited for the contexts of transitional justice. In the following section, I argue that rather than the Basic Principles influencing the incorporation of GNRs in TJ, it was TJ that inspired the content/contextual expansion of GNRs in the Basic Principles and IHRL.

TRANSITIONAL JUSTICE AND 'NEVER AGAIN'
On the GNRs timeline, it might be reasonable to assume that transitional justice comes into the picture after the LaGrand judgment, the ARSIWA and the Basic Principles, if for no other reason than the relatively short existence of the field. Nevertheless, this would be a false assumption, although it is important here to separate the legal term 'guarantees of non-recurrence' from the core values attached to GNRs in international law, that is, forward-looking measures that would prevent repetition. These core values were noted in the early developments of TJ (or 'justice in transition' at that time), which were shaped by the transitions from oppressive regimes across Latin America where the promise that violations will not be repeated enjoyed a prominent position. Notably, Argentina's National Commission on Disappeared People published a report entitled 'Nunca Más = Never Again' in 1984. At that time, the 'Never Again' phrase was associated with the advocacy carried out by human rights organizations and exiled groups to recognize the crimes perpetrated by the dictatorship. 47 In Argentina, as Crenzel reports, the phrase 'Never Again' itself became linked to the demands for not only truth but also (criminal) justice. 48 The Nunca Más report gained further importance for the public once it was offered and accepted as court evidence in the 1985 trials of military juntas. The truth commission model and its attachment to 'Never Again' was subsequently exported to various countries in Latin America (e.g., Guatemala, Chile, El Salvador). These developments shaped the emerging intervention of 'justice in transition' in Latin America, whose ultimate goal, whether through a truth commission or criminal trials, became to never experience such grave violations again.
It is in these developments, in Argentina and other mainly but not exclusively Latin American countries transitioning from authoritarian regimes to democracy in the 1980s, that Paige Arthur traces the conceptual origins of the field of TJ. 49 Arthur's arguments about the birth of TJ as a field distinct from human rights are crucial for understanding the travels and transformations of GNRs. Her study focuses on multiple actors who participated in the creation of TJ as a distinct and necessary response to transitions to democracy, providing an 'intellectual framework' for it. 50 Already at one of the first conferences on the topic of justice in transition, Jose Zalaquett, a Chilean lawyer and later a commissioner on the Chilean National Commission on Truth and Reconciliation (TRC), began to advocate that concrete measures such as state security reform should be undertaken for the reasons of 'guaranteeing to victims that violations would not be repeated.' 51 This is the language that at that time had only been voiced by the activists in Latin America ('Nunca Más') and the Human Rights Committee ('ensuring that violations are not repeated'). Until then, as I demonstrated in previous sections, the language of guaranteeing nonrepetition was reserved for interstate disputes, by and large practised as a right of states and not state subjects. Prioritizing legal and institutional reform, including security sector reforms Zalaquett talked about, was characteristic of the transfers of local-level advocacy against impunity to the international level as this process was marked by an overall focus on the political change (to democracy) rather than, for instance, social justice. 52 In Eastern Europe at that time, reforming and controlling the security sector by the new regimes became integral to the desired democratic and economic transformations of countries in transition. In broader international politics, 'good institutions' and 'good governance' emerged as most sought-after values in aid-receiving, developing states, defining how successes of state-and peacebuilding, and therefore sustainable peace, are measured. 53 While shaping the field's conceptual boundaries, the promoted measures in Latin America, well-placed within the global end-of-the-Cold-War context, gave normative content to the principle of 'Never Again' that was then formulated as an objective of the emerging field of transitional justice.
In 1990, around the time the conferences were taking place, Zalaquett published a seminal piece on confronting human rights violations. 54 This article was seminal for its influence on the setting of the intellectual framework for TJ, as noted by Arthur, but also for feeding content into the norm of non-recurrence in both TJ and IHRL, which has been omitted by TJ scholarship thus far. In the article, Zalaquett outlines nearly all of the measures today known as GNRs before any of the human rights documents mentioned in previous sections were drafted. He then saw 'dealing with transitional political situations' as 'a new area of human rights practice.' 55 Due to the myriad (at the time) unanswered questions such as what international legal responsibilities governments have regarding past human rights violations, Zalaquett found existing efforts 'inadequate.' 56 He, in turn, aspired to design a policy that would more appropriately deal with past human rights violations.
Such a policy would have two general objectives: 'to prevent the recurrence of such abuses; and . . . repair the damage they have caused.' 57 Here we see a clear separation between reparations and non-recurrence, a topic that caused a great deal of disagreement among the ARSIWA drafters in the subsequent decade. In addition to 51  The Law of 'Never Again' 11 measures of reparation, punishment and clemency, Zalaquett gave great weight to preventive measures of non-repetition using examples of domestic practices in Guatemala, the Philippines, Spain and Panama. Admitting that 'there exists no single formula' for how best to prevent human rights violations, Zalaquett outlined the following 'new' measures of state, institutional and legal reforms. Firstly, he spoke of reinforcement of the legal system's protection of human rights (e.g., amending constitutions and ratifying human rights treaties) and secondly, of reforming/ strengthening institutions to reinforce the protection of human rights (judicial reform, human rights education and dismantling of intelligence services or other units responsible for abuses). 58 In the year after the article was published, Zalaquett sat on the Chilean TRC. The commission's lengthy final report includes a set of recommendations regarding the prevention of human rights violations in the country, in line with its mandate. 59 In the introductory remarks written for the English language version of the report, Zalaquett repeated the two ultimate objectives of a 'justice in transition' policy from his 1990 article, with preventing the repetition of atrocities being one of them. In Zalaquett's words, the Commission's mission was to decide how such an objective could be achieved. The prevention recommendations stress the 'significant flaws and shortcomings' of the domestic legal system which disable it to effectively protect human rights. 60 To that end, the Commission recommended a series of measures of institutional and legal reform: aligning Chile's legal framework with international human rights law, reforming the judicial branch (including legal training), ensuring that the armed forces, security forces and police exercise their functions in complete accordance with the obligation to respect human rights, creating an institution to protect human rights and making changes in the legal order in constitutional, criminal and procedural matters to better protect human rights. These elaborate measures are an extension to the 1990 article by Zalaquett, although ideologically identical.
These recommendations quickly became a trend. In the years to follow, truth commissions generally adhered to non-recurrence as an objective of transitional justice mechanisms in their recommendations. In the Guatemalan peace process, for instance, avoiding repetition of violent events was connected to the right to truth, manifested in the establishment and work of the Clarification Commission. 61 Furthermore, in its final report, the truth commission in El Salvador recommended measures of institutional reform to prevent the repetition of violent acts, broadly categorizing them as relating to justice, protection of human rights and reconstruction of the national civil police. 62 58  Zalaquett's interpretation of these various measures of institutional and legal reform as measures best preventing human rights violations, and the subsequent practices of justice in transition, helped give content to the otherwise content-ambiguous norm of non-recurrence in IHRL. An important yet insufficient argument would point to the stark textual resemblances between the measures recommended by the Chilean commission and those included in the Basic Principles, albeit stripped of context. 63 More notably, both the Chilean Commission and Zalaquett himself were cited in van Boven's appendix to the 1993 draft Basic Principles. 64 Nearly three decades later, Prof van Boven observed that the 'certainly large' influence of the Latin American experience is manifested not only in the Basic Principles but in its contemporary, the Joinet/Orentlicher principles on impunity. 65 In her 2005 report, Dianne Orentlicher notes that '[s]ome aspects of' listed GNRs are 'characteristic of periods of restoration or transition to democracy and/or peace that prevailed in Latin America and other regions at the time the Principles were drafted,' of which Chile was a representative state. 66 Supporting this is the pivotal role Chile played in the creation of the Basic Principles, as the drafting consultations were chaired by Chilean ambassador Alejandro Salinas. 67 Finally, the overall political backing for these principles came from Latin American countries emerging from oppressive regimes. 68 It is here where I make the first and core claim of the early TJ developments' influence on the norm of non-recurrence in IHRL.
Until the Basic Principles were issued, however, the understandings of measures most adequate for ensuring 'Never Again' in TJ were not explicitly phrased as 'guarantees of non-recurrence' but rather as prevention of repetition. Expectedly, a survey of English-language newspaper articles published in the 1980s and 1990s I conducted resulted in notable references to GNRs only regarding solving interstate disputes. The language of GNRs was repeatedly used in the Afghan crisis on behalf of the Afghan government to, for instance, seek appropriate political guarantees from the Soviet Union against military invasions and other internal affairs interferences. In the following decade, GNRs seldom made appearances in state officials' statements regarding conflicts in Iraq and Lebanon and generally maintained a rather low profile.
In the early 2000s GNRs experience two distinct 'booms' in references -one in discussions about state responsibility in connection to the LaGrand judgment and the ARSIWA, and the other in relation to the developments of the early TJ scholarship and the Basic Principles. In the former, GNRs were discussed as reviewed above, more as a debate about state obligation than about norm content. On the latter front, GNRs become attached to, for example, history education about the stolen children in Australia in an article Durbach and Kirk wrote for the Australian on 14 August 2000, and restitution for the crimes committed against 'comfort women' in Japan, as demonstrated in a contribution to CNN by Shinsho and Kamimura on 12 December 2000. Yet as the drafting of the Basic Principles concluded in 2005, first references to the Principles and GNRs as legal and institutional measures of reparation were made in Sierra Leone concerning the establishment of the Special War Crimes Court, 69 in Nigeria regarding widespread forced evictions 70 and in a handful of other cases of human rights abuses in India, Algeria and the US. In all these cases, GNRs and the other four types of reparations, as per the Basic Principles, were called upon by Amnesty International, Zalaquett's former employer. Present-day references to the five distinct types of reparations, GNRs being one of them, or to GNRs in isolation, are too numerous to list.
The key aspects of norm transformation in and by TJ in this stage are twofold. Firstly, GNRs were given content and context expansion based on early TJ practices and scholarship from Latin America. Secondly, the 'Never Again' promise, inseparable from justice in transition, was increasingly becoming equated with a list of institutional and legal reform measures, those that originally stemmed from Chile. Both aspects of transformation were contingent on the growth of TJ as a legalized field, blending in the post-Cold War politics of regime change and the overwhelming emphasis on good governance and the rule of law. In the formulation of TJ and the rule of law as 'mutually reinforcing phenomena,' 71 institutional reform measures appear crucial for restarting the rule of law in the transitioning society. This is because the legitimacy of this revived rule of law depends both on the content of laws and the process of making them, which necessitated reformative processes such as vetting. 72 This stage of norm transformation therefore demonstrates transfers of local-level advocacy to the international arena and subsequent decontextualization of the norm content from Chile/Latin American-specific contexts to a staple of universally applicable reparations.

THE UNITED NATIONS AND FURTHER NORMATIVE EXPANSIONS
Equating the initial ideas about future-oriented actions aimed at fulfilling the 'Never Again' promise to GNRs has been fortified through to work of the United Nations (UN) and its intervention in TJ. Since 2004, transitional justice has been high on the UN's agenda. In the first major TJ report, Secretary-General Kofi Annan wrote that it was the horrendous events of the 1990s that showed that there can be no immediate nor long-term peace without appropriate redress for grievances and fair administration of justice. 73 Transitional justice was thereby becoming a norm. In 2010, the Secretary-General published a guidance note on TJ in which the duty of prevention, from which GNRs stem, was discussed. 74 Importantly, the report highlights that TJ processes should 'ensure the right of victims to reparations, the right of victims and societies to know the truth about violations, and guarantees of non-recurrence of violations, in accordance with international law.' 75 As to the content of GNRs, this guidance note aligned with the measures listed in the Basic Principles.
The UN agenda on TJ, of course, culminated with the establishment of a Special Rapporteur on the promotion of truth, justice, reparation and guarantees of nonrepetition by the Human Rights Council in 2011. 76 The Special Rapporteur is widely regarded as the 'transitional justice' rapporteur and truth, justice, reparations and guarantees of non-repetition as the four pillars of the field. Without any explanations, guarantees of non-repetition -and not, for instance, prevention -have been, firstly, made a pillar of transitional justice, and not an objective, and secondly, separated from reparations, contrary to the Basic Principles.
The Special Rapporteur was given a lengthy set of tasks, including recommending how the elements of the mandate, such as GNRs, can be improved and strengthened, and building their normative framework, which the first Special Rapporteur Pablo de Greiff did in three separate reports. In the first and most elaborate report, De Greiff provides clarifications to certain conceptual ambiguities regarding GNRs. 77 He defines a guarantee of non-repetition as a function performed through a variety of measures, to which previously victimized societies as a whole are entitled, and where state institutions are duty bearers. In that sense, guaranteeing non-repetition is no longer a moral promise but 'an object of rational policymaking.' 78 The GNRs in TJ, unlike in PIL, are not exceptional but normative; always specific and not general guarantees, concrete interventions that oblige institutions of the state and jointly contribute to diminishing the likelihood of repeated violence. 79 As the legal foundations for GNRs, the report relied on the developments of GNRs in IHRL, including the work of the Human Rights Committee and regional human rights courts. Therefore, the context in which these kinds of GNRs become applicable is in line with the Basic Principles -gross violations of human rights and humanitarian law.
The norm transformation undertaken by the Special Rapporteur in this report is not one of context but of content. In De Greiff's theorization, GNRs incorporate multiple measures that are feasible and could be achieved through TJ measures which require overarching mobilization and benefit the entire previously victimized society. In his view, once the basic preconditions for institutional interventions such as ensuring security and access to legal identity for all are met, there exist three fields of intervention through which policies relating to GNRs should be structured. 80 The first one is the known level of official state institutions, inclusive of legal, judicial and constitutional reforms, while the next two are the level of civil society and of cultural and personal dispositions that, inter alia, includes educational reform and trauma counselling. For each of the interventions, a series of measures were proposed, some of which had never appeared in any documents or notable scholarly writings on GNRs before.
The Special Rapporteur's report on GNRs resonates with the initial understandings of preventing future violations shown in Zalaquett's writings and the Chilean context. While the Chilean TRC emphasized that measures of legal and institutional reform do not suffice to ensure non-repetition and asserted that human rights must be included in formal education and manifested through symbolic measures and the attainment of truth and justice, these recommendations were not replicated in the Basic Principles. They were also not fully included in the early TJ scholarship which preoccupied itself with state structures and laws. Like the TRC, the Special Rapporteur discusses education and the changing of the culture of human rights as conditions needed to prevent future atrocities. A complex function such as non-recurrence, the work of the Special Rapporteur similarly posits, could not be limited to measures of legal and institutional reform. Although this expands beyond IHRL, it should not be seen as problematic but as a doctrinal attempt to capture the complexities of the non-repetition of mass violence.
In his second report on GNRs, De Greiff discusses security sector reforms, in particular vetting and its contributions to non-recurrence. 81 He believes that vetting can be an enabling condition for transitional justice, meaning that TJ institutions will work better if the institutions responsible for abuses are vetted. Finally, in the 2017 examination of regional approaches and advances to GNRs, De Greiff admits that GNRs are rarely used outside UN circles and that even within the organization few documents utilize the term. 82 De Greiff is right in saying that only if we forget about the GNRs which are operational in PIL. His criticism is that transitional justice has not been especially good at taking advantage of this new knowledge of the norm, which is why GNRs remain its least developed aspect.
De Greiff's successor, Fabian Salvioli, has also addressed GNRs in his reports, mainly reiterating De Greiff's writings yet manifesting some conceptual confusion. 83 In a report on domestic reparations, Salvioli argues that the right to reparation is important as a GNR but also references the IACtHR case law which lists GNRs as a form of reparation. 84 Such references could imply reparative as well as future- oriented characteristics of the concept in TJ. In yet another joint study, it was reiterated that GNRs are doctrinally the least developed pillar, but that a 'vast amount of knowledge and expertise on the topic, albeit fragmented,' exist, citing the three reports on GNRs by De Greiff reviewed above which, as explained, are not firmly empirically grounded. 85 Nonetheless, such phrasing signifies, in my view, that these three interventions on GNRs by the Special Rapporteur have become 'the knowledge' about GNRs in transitional justice. Such knowledge is thus far the final and most expansive transformation of the norm of non-recurrence. This new 'knowledge' is of course in tension with the previously detected attempts to equate GNRs with broadly defined measures of legal and institutional reform. A brief review of selected international TJ advocacy provides insights into whether and how this further normative expansion of the norm of non-recurrence has been implemented in practice. Looking at Amnesty International, for instance, the number of references to GNRs increases dramatically post 2011, yet these typically introduce GNRs as an aspect of 'full reparation' or an element of comprehensive TJ. 86 In the case of Sri Lanka, towards which much of Amnesty's TJ advocacy has been directed in the past decade, the language of GNRs as something drastically different or expanding beyond legal and institutional reform is not noted. In recommending the necessary GNRs for the Sri Lankan government to undertake, the organization cites the Basic Principles 19-23 87 or otherwise elaborates on more context-specific legal and institutional measures such as witness and victim protection or an enforced disappearances act. 88 At the International Center for Transitional Justice (ICTJ), although there is evidence of the Special Rapporteur's normative expansion taking root, some level of conceptual ambiguity is maintained. The information available on the organization's website indicates that GNRs are not a standalone category of research and practice, but it is rather 'institutional reform' that appears as 'the fourth pillar' of TJ, next to truth and memory, criminal justice and reparations. A further review of the ICTJ's publications leads to a conclusion that while the language of preventing repetition is often inseparable from recommended or applied measures of institutional reform, certain practices conceptualize GNRs as beyond institutional reform. 89 In a recent report published by the Working Group on TJ and SDG16+, which is convened by the ICTJ but also includes organizations such as Impunity Watch and Swisspeace, it is De Greiff's conceptualization of GNRs that was advocated for. One of the recommendations issued was to '[e]xpand the understanding of means to achieve guarantees of nonrecurrence to include . . . civil society, faith-based, cultural, and individual-level interventions.' 90 The Center also runs an annual course on prevention and GNRs and the role of TJ in Barcelona which has accommodated some of the noted conceptual expansion and confusion over the years. For instance, in 2019, the overview of the course on the ICTJ's website acknowledged that GNRs are 'defined beyond the institutional sphere,' while in 2020, GNRs emerged as 'under-explored' and how TJ prevents recurrence of violence is 'less clear, and more contested.' Nevertheless, the course still promises to explore TJ practices and their contributions to avoiding the recurrence of violence across numerous case studies. Overwhelmingly, the ICTJ's treatment of GNRs reads as strategically malleable enough, both universally applicable and ambiguous, to be adjusted to any new contexts and situations that may arise. 91 REVOLUTION OR CONFUSION? I have argued that transitional justice practice and scholarship have aided the transformation of the content and context of GNRs in international law but that more recently GNRs have been distinguished as qualitatively different in TJ as opposed to IHRL. GNRs in TJ are a prime example of what Christine Bell has usefully framed as 'the new law' of TJ in her analysis of TJ's mingling with international legal norms on accountability and the process of regime merge. 92 Throughout the article, I evidenced how illustrative GNRs are of this regime merge or, perhaps more adequately phrased, regime blur. As Bell concludes, the developments stemming from the regime merge can be read as either positive or negative; they can be welcomed for their normative flexibility or considered dangerous for 'exceptional application of international law.' 93 The story of the norm of non-recurrence and its relationship with TJ is too unexplored in practice to be proclaimed as either outright positive or negative. What this article brings to the forefront is a tension between normative revolutions and confusions triggered by TJ which can and will have significant effects on local TJ practices relating to 'Never Again.' In the preceding sections, I have identified the decontextualization of the norm content from Chile and Latin America and its transfer to universal applicability. In this process of decontextualization, the 'Never Again' promise was equated to the international legal norm of GNRs and GNRs, as a pillar of TJ, to institutional reform. It is possible that due to the widespread application of TJ as a project, 94 this equation is prevalent in domestic policymaking. As Arthur asserts, the norm entrepreneurship in transitional justice has involved attempts to produce a body of knowledge about causal relationships between justice measures and transitions, in all their diversity, 95 a trend that is characteristic for policy-oriented thinking.
The worry is that the technocratic current in TJ has already established a causal relationship between 'Never Again' and institutional reform across contexts, which are much different from those in which this equation was initially made. In addition to Amnesty International's treatment of GNRs, a further illustration would be the 2015 European Union's Policy Framework on support to TJ which lists the fourth pillar of TJ as 'guarantees of non-recurrence/institutional reform' thereby making the two equal and inseparable. 96 The danger of 'institutional fixes,' of tying legal and institutional measures to the value of preventing renewed violence, as these two examples suggest is in that they could limit further productive thinking about 'Never Again.' 97 The norm transfer and transformation could have therein normatively limited GNRs in TJ projects (as opposed to TJ practice at large), omitting measures that could prove more effective in the maintenance of lasting peace, or excluding responsibility-bearing actors other than the state. If this equation implies that there is a causal relationship between institutional reform and non-recurrence and that non-recurrence can hence be measured through the application of institutional reform measures, then the utility of the norm of non-recurrence across contexts ought to be questioned. Would its utility be more enhanced with its original characteristics found in the PIL regime, where the norm is flexible, exceptional and even ambiguous, than when it is colonized by a broad category of institutional reform? What is certain, nevertheless, is that the understanding that guaranteeing non-recurrence is the right of individuals and not only states is a transformation indebted to the birth and development of transitional justice. Based on this spread of the sense of both non-exceptional state obligation and individual right to non-recurrence in postconflict contexts, transitional justice has aided a significant norm transformation, one might even say a revolution.
On the opposite end is the possibility that the ongoing conceptual expansion by the UN and strategic maintenance of norm malleability, as evidenced by the ICTJ's course on prevention and GNRs, allow for the flexible characteristics of GNRs to persist to the extent that selective implementation of the norm is accommodated. Making GNRs qualitatively different in TJ than in IHRL opens space for recontextualization of the seemingly universally applicable equation of GNRs to institutional reform. A good example of this is the 2016 Final Agreement between the Colombian government and the FARC. 98 The agreement uses the term 'guarantees of non-repetition' dozens of times and embodies the principle of non-repetition throughout the proposed TJ intervention, therein prescribing numerous measures that need to be established to reach the objective of non-recurrence. These various measures are inclusive of rural reform, accountability and political participation, and victim-oriented measures such as truth recovery, national collective reparations and even strengthening the mechanisms for human rights education. The current prospects for materializing these measures aside, the agreement demonstrates a context-based approach to GNRs conceptualization, promoting non-recurrence as overall implementation of all provisions in the agreement. What is missing from this comprehensive agreement, however, are measures of institutional reform such as security sector reform and vetting that would otherwise likely be triggered in IHRL. The Colombian peace agreement, one could argue, is thus a curious case of either welcome context sensitivity emerging out of political negotiation, or random selectivity of measures phrased as GNRs in the absence of clear guidance on specific GNRs stemming from the principles of state responsibility.
Expanding the conceptual realm of GNRs presumably leaves more room for context specificity; yet detachment from international law, as seen in the Colombian case, also creates more opportunities for non-applicability and non-application, to the disadvantage of prospective beneficiaries. The confusion regarding GNRs is thus not only normative and conceptual but also concerning legal obligations and their practical delivery at home. The flexibility of the norm, while characteristic of GNRs in international law, is contrary to the non-exceptional nature of 'Never Again' in TJ. Both the normative expansion and maintenance of openness in TJ have made GNRs more necessary than international law proposed they should be, to the point where they are universally needed and applicable. Hence the usefulness of the concept is that in any peace processes incorporating TJ, like in Colombia, there is a sense of necessity to use the language of GNRs to demonstrate a sense of duty to indeed ensure non-repetition of mass violence through specific actions. Yet, this seemingly welcome marker of context specificity is also inseparable from the malleable character of TJ itself 99 and therefore the emergence of GNRs as purposefully mandatory yet still conceptually ambiguous. The continuous normative expansion and openness keep the norm-affirming potential of GNRs weak and could plausibly enable 'Never Again' to become whatever is easiest, and not whatever is most necessary to implement. To put it bluntly, it would mean drawing from not multiple but no legal regimes whatsoever and leading the 'too ambitious, too quickly' approach to norm transformation to result in inaction.
If this were the case, it would mean that the 'new law' of TJ is more concerned with making the 'Never Again' promise than with fulfilling it, and as such embodies a pillar of mere verbal assurances rather than meaningful and revolutionary GNRs. A step forward should not be to entirely reinvent GNRs in TJ; it is a slippery but highly phenomenal concept TJ has brought into IHRL. Instead, application of GNRs should mean allowing the lived experiences of beneficiaries to be reflected in the norm content, which making GNRs normatively distinct in TJ could do to some extent, while still tying the concept to its legal origins and the principles of state responsibility to guarantee non-recurrence in IHRL to ensure that any political abuse of the obligation is minimized.

CONCLUSION
In this article I have shown how transitional justice has transformed both the contents and contexts of the principle of non-recurrence and has furthermore helped stretch the principle of state responsibility in the human rights regime. Until the emergence of TJ as a distinct field and the subsequent developments in IHRL with 99 See Rowen, supra n 91. regards to victims' rights, assurances and GNRs suffered from non-uniform state practice and were known for their exceptional character. Invoking the principle of non-recurrence most frequently meant a mere verbal commitment, an apology and a promise not to act in a violating way again. However, the contexts of transitional justice bring mass violations into the picture, violations perpetrated by state agents towards their citizens and enabled through various state structures, including laws and institutions. The influence of the developments in Chile and other Latin American countries on the Basic Principles and the establishment of GNRs as a form of reparation which such violating states could be entitled to provide is, without a doubt, a welcome and most remarkable development. Still, I have also provided empirical examples that further complicate the conundrum about non-recurrence in TJ, demonstrating that the norm of non-recurrence has been both equated to institutional reform measures and simultaneously normatively expanded and maintained as sufficiently ambiguous to be applicable and necessary across varying contexts. The story of GNRs is therefore one of decontextualization of local understandings of the norm and further attempts to recontextualize the same norm in light of or despite the content given by TJ. The effects of these processes on both actual non-recurrence and perceptions of it in local contexts, if any, will have to be studied empirically while building on the legal and conceptual foundations and tensions established in this article.
The Law of 'Never Again' 21