The International Criminal Court’s Selectivity and Procedural Justice


 Prosecution selectivity is one of the most intractable dilemmas in international criminal justice. It is of little surprise, then, that the International Criminal Court’s (ICC) selection of cases has long been subject to critical debate. This article contributes to the literature by analysing the ICC’s selection procedure from the perspective of affected communities. Vis-à-vis this target audience, the article critiques the procedure’s effectiveness against a measure of perceived legitimacy. Using a Rawlsian model of imperfect procedural justice, the analysis explains the specific shortcomings of the Office of the Prosecutor’s (OTP) selection procedure in being sufficiently consistent, impartial and representative. In turn, this lack of procedural fairness may reduce the likelihood that the OTP selections are perceived as legitimate within affected communities. More broadly, the article argues that the OTP is unable to reach the ‘fairest’ possible prosecutorial decisions as to situations or cases — culminating in the conclusion that its selection procedure makes a limited (if any) contribution to the Court’s perceived legitimacy. The article triggers reflection on the Court’s relationship with target audiences and concludes by making practical recommendations directed at improving the OTP’s selection procedure.


Introduction
The International Criminal Court's (ICC) Office of the Prosecutor (OTP) is the Court's 'engine-room'. Its investigations and prosecutions are pivotal to the operation and ultimate effectiveness of the Court. The OTP's primary focus is on prosecuting cases, i.e. seeking convictions, but -in the process -it is also concerned with influencing the perceptions of target audiences. 1 It is not only desirable, but necessary, that the OTP gains the support of domestic audiences such as governments, civil society, victims and communities affected by the commission of crimes. 2 Such support is initially gained by the OTP's selection of cases, something that makes the most enduring contribution to the Court's perceived legitimacy -a particular audience's acceptance of its authority. 3 The OTP's selections and the Court's perceived legitimacy are inextricably linked. The relationship can best be described as mutually reinforcing: selections have the potential to either enhance or diminish the Court's perceived legitimacy, with the latter shaping the extent to which those selections can, in fact, do so. For over a decade those prosecutorial selections have experienced a decline in public confidence -particularly on the African continent. The general trend of criticism has focused on the fact that investigations have almost exclusively targeted nationals of African states. 4 There is also evidence of an intra-situation pattern whereby selections defer to the interests of the state and thus alleged crimes committed by Government forces are overlooked, such as in the Situation in Uganda and in the Situation in the Côte d'Ivoire. 5 A 2017 African Union strategy aptly surmised the consequences of prosecution selectivity by declaring the Court is 'riddled with . . . struggles over its perceived legitimacy'. 6 Against this background, this article examines the OTP's procedure in selecting cases for investigation. The concept of procedure is understood here as the method followed when selecting a case to be prosecuted, and this method includes prosecutorial discretion. 7 The article's precise focus is procedure from the perspective of affected communities -including victims and those generally affected by the commission of crimes (e.g. those having witnessed crimes and/or having suffered material, physical or psychological harm in the aftermath). 8 To date, scholarly literature has tended to focus on the legitimacy and transparency of prosecutorial discretion. For instance, several accounts have considered the case for prosecutorial guidelines to help illuminate the 2 Ibid. exercise of discretion. 9 Other accounts have offered comparative reflections on prosecutorial discretion operating within international criminal tribunals and national courts. 10 Most of all, the literature has tended to critique the politicized nature of prosecutorial discretion and the production of problematic selection patterns. 11 The latter research includes warnings about the risk of prosecutors adopting selection practices that are based on the alleged perpetrator's group identity, e.g. their ethnicity, nationality or political affiliation. 12 However, there has been little scholarly attention for the selection procedure from the viewpoint of its effect on and perception by affected communities. 13 This research gap is surprising and significant for, at least, two reasons. First, an analysis of the Prosecutor's selection procedure helps to illuminate the Court's 'target audience' dilemma. The Court lacks a single target audience, i.e. a defined constituency to whom it is primarily accountable and responsive. 14 The Court is expected to 'speak' simultaneously to various audiences, ranging from those that are defined by abstract concepts (such as the 'international community') to those of a more concrete nature (such as potential perpetrators, states parties, civil society, donors, victims and crimeaffected populations in general). 15 This lack of specificity also contributes to uncertainty about the Court's goals. The Court is expected to contribute to a range of often conflicting goals, e.g. deterrence, peace, reconciliation -with ample disagreement as to which goals are to be privileged, and as to when and whether those goals should reflect global and/or local priorities. 16 The said uncertainty invests several of the Court's activities including, for instance, prosecutorial decisions: it is not clear, inter alia, whether they should be taken by giving 'priority to the needs for redress of affected local communities or focus[ing] instead on giving voice to particular norms of the international community'. 17 In that regard, the present analysis explains that the Court's selection procedure pays only limited attention to its perceived legitimacy in affected communities.
Secondly, the lack of scholarly attention towards the Prosecutor's selection procedure is surprising because such procedure is so fundamentally connected to the interests of affected communities. The OTP's selections are -at least in part -made on their behalf and for their benefit; in the words of Fatou Bensouda, to prosecute is to 'stand up for the victims and affected communities'. 18 However, prosecutorial decisions are often made opaque because The Hague is thousands of miles away, physically and morally remote from affected communities. Thus, there is a convincing case, in the words of Goldston, to 'bridge the yawning gap between The Hague-based Court and [affected communities] across the world . . . [and balance] their hopes for justice against their often-uncertain knowledge of the Court's operations and limitations'. 19 Of course, the Court's communication and outreach strategies help to explain the OTP's selections but this masks the fact that the procedure upon which those selections rest should inherently be satisfactory for those that are most concerned by its outcomes. 20 From this perspective, the analysis reveals that the OTP's selection procedure generates limited support among affected communities.
This article takes research on selectivity in a new direction by adopting the perspective of procedural justice. Hitherto, procedural justice has mostly been considered within domestic criminal justice contexts, particularly in respect of the procedures followed by enforcement agencies and other public institutions. 21 Thus, procedural justice may provide an original heuristic device to examine case selections within the context of international criminal justice. This does not mean that one can ignore the undeniable differences between domestic and international criminal justice. Rather, the ensuing analysis may help to uncover whether international criminal procedure faces inevitable and inherent limits in terms of procedural justice. Before turning to procedural justice, however, it is imperative to define the notion of perceived legitimacy.

Perceived Legitimacy
Perceived legitimacy is concerned with an audience's subjective belief(s) in an institution's right to rule. 22 It is synonymous with sociological accounts of legitimacy. In this sense, legitimate institutions are those that are perceived as desirable, proper and, when appropriate, correct in exerting their influence and power. 23 These accounts are distinguishable from normative accounts of legitimacy, which tend to be focused on the legality or probity of institutional decisions or procedures. 24 Of course, an institution's adherence to legality may be a source of its perceived legitimacy, but the two are not necessarily interchangeable. In fact, the public may deem an institution legitimate for reasons that may seem unfair or arbitrary and against the rule of law. 25 To put it simply, the Court's perceived legitimacy refers, exclusively, to the audience's acceptance of its authority to rule and judge disputes. 26 Assessments about the degree of perceived legitimacy require, a priori, identification of a specific audience. As mentioned above, the Court 'speaks' to multiple constituencies simultaneously, but among those constituencies is a crucial one, i.e. 'affected communities' -generally comprised of victims and those most affected by the commission of the crimes. 27 At least in part, the OTP's selections are made in the pursuit of delivering justice to affected communities. However, justice does not exist in the abstract -it must be perceived or seen to be done. 28 And so, the Court must first be perceived to be legitimate if justice -whenever it comes -is to be seen by affected communities as having been done. 29 One can readily cast this requirement in terms of 'effectiveness', i.e. the extent to which selections boost the Court's perceived legitimacy in those communities. 30 The perception of the Court's legitimacy by affected communities (and indeed by any community) is a complex, multi-layered and psychological phenomenon. The very concept of perception can be understood in two ways: first, as the ability to see, hear or become aware of something, principally by one's senses; and, secondly, as the way something is regarded or understood. 31 These meanings are of course related, because to have a perception is, first, a psychological process by which something from one's environment is interpreted and, secondly, this initial perception shapes how that something is regarded or understood. In this sense, it has been argued that perceptions encompass two major epistemological points of view: the objective and the subjective, with the former being a material reality that people see, and the latter being a personal reality specific to each and every individual. 32 From an objective perspective, perceptions of the Court are likely to be informed by various material attributes, e.g. its proximity, the extent to which participation is permitted, the degree of access to information about the Court and the type or quality of justice it is seen to dispense (in contrast to, say, alternative and possibly local mechanisms to cope with mass atrocities). These attributes can, individually or in combination, delegitimize the Court in the eyes of affected communities. From a subjective perspective, no single affected community is an empty container, but it is comprised of diverse individuals whose own perceptions are likely to be influenced by their own socially conditioned beliefs and convictions. 33 Individuals are shaped by their 'anchors' (e.g. ethnic, political, religious or social affiliations) which produce cognitive and emotional biases; these have the effect of first shaping and then hardening individual perceptions, and hence making them difficult to change. 34 In summary, both objective and subjective factors are crucial in understanding perceptions of the Court.
What, then, does the existence of such subjective factors mean for the Court's perceived legitimacy? First, as Milanović persuasively argues, subjective factors enable one to be 'realistic about the causal factors that drive public perceptions of the work of international criminal tribunals'. 35 Subjective factors are always, inevitably, likely to play a significant part in the formation of perceptions, no matter the extent to which the Court tackles some of its objective limitations. In that respect, one needs to be intellectually honest about the causal factors that contribute to public perceptions of the Court, without necessarily being too cynical about the role the Court can play. 36  those that are entrenched in their scepticism, has always been inherent in the very project of international criminal justice. 37 The Court's orientation has, and must always be, to increase its legitimacy because pessimism or scepticism -all too prevailing and exacerbated by political elites espousing anti-Court sentiment -is not something that can be ignored.
There needs to be, then, a more nuanced and sophisticated account of the Court's perceived legitimacy. This account should, first, distinguish the potential objective and subjective factors that form individual perceptions; and, secondly, explore how changes in objective factors can influence the impact of subjective factors. In this regard, long-term perceived legitimacy is what really matters: 'diffuse support' -a reasonable and stable recognition of an institution (i.e. the Court) as legitimate, coupled with a general willingness to accept its decisions. 38 This type of support can be distinguished from specific support, i.e. a positive attitude towards (and/or approval of) particular institutional decisions or policies. 39 As Baird elaborates in more detail, [d]iffuse support is the belief that . . . the institution itself ought to be maintained . . . trusted and granted its full set of powers. [Its development] suggests that people maintain a 'running tally' that increases over time with pleasing policy decisions. Over time, the running tally develops into a reservoir of good will that serves to insulate support from later disagreeable decisions.' Satisfaction with particular decisions, though at one time the primary source of diffuse support, become over time, separable from a willingness to support an institution. One can be dissatisfied with a recent decision and yet maintain a relatively strong level of diffuse support. 40 The OTP's initial contribution to diffuse support is its selection of cases, particularly the choice of those accused. I concede that -in divided societies having endured violence motivated by ethnic, religious or political reasonsthe impact of prosecutorial selections on perceived legitimacy may be a zerosum game, i.e. they may attract support in some affected communities whilst simultaneously triggering antipathy in others. However, it is precisely such zero-sum games that diffuse support seeks to mitigate as, over a period of time, communities' support will become detachable from specific prosecutorial decisions.

Prosecution Selectivity and Procedural Justice
According to Thirlway, 'procedure, by definition, is no more than a way of getting somewhere'. 41 It follows that procedural justice is getting somewhere that is just, or a procedure calculated to produce a just decision. 42 Considerations of distributive (or substantive) justice help to determine whether those decisions are, in fact, just. This determination is often based on its fairness and is commonly assessed against the division of burdens, punishments, benefits, rewards or shares in society. 43 By contrast, procedural justice is concerned with whether the procedure is fair. 44 In legal contexts, such fairness is based on how norms, principles or rules are applied in any given case (because the law generally aims to achieve outcomes through its application). 45 Therefore, procedural justice requires procedural fairness, with procedure and decision being linked because, broadly speaking, the fairer the procedure, the fairer the eventual decision. 46 However, this relationship begs two essential questions: (i) do procedures that lead to fair decisions exist, and (ii) can one know whether a decision is itself fair?
These questions were contemplated by Rawls in The Theory of Justice. He distinguished three types of procedural justice: perfect, imperfect and pure. 47 Perfect procedural justice is rare and occurs when an independent standard can help determine whether a decision is fair and a procedure exists that is guaranteed to produce one. Imperfect procedural justice is where there is an independent standard to help determine whether a decision is fair, but there is no feasible procedure that can be sure to lead to such an end. Finally, pure procedural justice is where there is no independent standard to determine a fair decision, but by following the fairest procedure, one will produce a 'correct' 48 decision, whatever it happens to be. This final type of procedural justice leads to a decision that is fair by virtue of scrupulously observing an infallible procedure. 49 These three accounts are useful in specifying the current enquiry. To begin, the present question is not about the fairness of the procedure's results, i.e. the selections themselves. Instead, the starting premise is that those selections can be improved and, thus, to enquire into the procedure that produces them. This article aims to optimize the potential of this selection procedure -an aim that finds a degree of expression in Rawls' imperfect procedural justice. Admittedly, even on Rawlsian terms, whether an independent standard in fact exists to determine the fairness of final selections -let alone one that could draw sufficient consensus if one were to exist -is incommensurable. Nonetheless, the approach of imperfect procedural justice retains humility about procedures and accepts that one can never be completely sure that any procedure will lead to the fairest result at any given time.
This perspective is in stark contrast to the OTP's frequent rhetorical claims that its procedures are inherently faultless and, ipso facto, always lead to the fairest selection. The OTP has long amplified its faith in pure procedural justice by persistently claiming that its selection procedure is based on a strict legalist approach. By this, the OTP maintains that its selections are exclusively based on a black and white technical application of the Rome Statute, with little to no space for the exercise of any discretion, political or otherwise. 50 However, these claims entirely unravel when one considers the actual configuration of the OTP's selection procedure.
In reality, indeed, the OTP's selection procedure includes a dose of discretionary decision-making. Discretion by nature is uncertain because, even though decision-makers are asked to implement rules, they are given significant leeway as to their application. The exercise of discretion is therefore not a mechanical process but, requires subjective human judgment. 51 A few considerations illustrate the extent of such discretion. First, it is well known that the OTP selects situations after a state referral to the Court, a United Nations Security Council (UNSC) referral in accordance with Chapter VII of the UN Charter or after exercising its proprio motu powers. 52 Already at this early stage, the OTP exercises discretion by deciding whether to open an investigation on the basis of those referrals or by selecting those situations that are within the Court's jurisdiction, are of sufficient gravity, etc. Secondly, as governed inter alia by Article 53(1)(a)-(c) of the Rome Statute, the Prosecutor exercises discretion by considering admissibility (including the tests of 'complementarity' and 'gravity') and the 'the interests of justice'. 53  Prosecution Selectivity and Procedural Justice 9 of 33 decisions are shaped by opaque factors such as the extent to which a state has demonstrated sufficient willingness to investigate or the extent to which a prosecution would be in the interests of justice. 54 In addition, the OTP must assess whether the final choice of accused includes those who hold 'the greatest responsibility' for the most serious crimes -and this choice also requires reconciling available evidence, enforcement capability and other prosaic questions such as how the limited resources available to the Office should be managed. 55 In summary, the driving force of the selection procedure rests on the unpredictable exercise of prosecutorial discretion. In this light, the question for the present analysis is: what makes this selection procedure fair?
Before one can answer this question, one must consider the general components of procedural fairness. Researchers began to meaningfully discuss such questions in the 1970s and 1980s. First, Thibaut and Walker suggested that fairness is demonstrated when those affected can influence the procedure and thus exert a degree of control over the eventual decision. 56 Later, Leventhal speculated that fairness required procedure to demonstrate six components: (i) consistency across circumstances, persons and over time; (ii) impartiality and/ or maintaining a suppression of bias in a key decision-maker; (iii) accuracy, i.e. being based on a full range of reliable information; (iv) 'correctability', i.e. subjection to a correction mechanism or an appeal of the final decision; (v) representation of the interests of groups affected by the procedure; and (vi) ethicality, i.e. conformity with commonly held moral values. 57 These principles paint procedural fairness as an objective or a normative question -simply based on complying with standards and safeguards. 58 However, much of the literature interprets procedural fairness to be a subjective and psychological response, i.e. something based on the perception of those affected or involved, like disputants (defendants or victims). 59 Objective principles retain their significance only insofar as they establish benchmarks that are indicative of what individuals are likely to accept as a fair procedure.
Research suggests that perceptions of procedural fairness are psychologically distinct from perceptions of the fairness of eventual decisions. 60 It is precisely because of this separation that individuals are more well-disposed towards generally describes duties to act fairly. Natural justice has become recognized within an individual's procedural rights e.g. the right to a fair hearing and the right to representation. In common law legal systems, these procedural rights have often formed the basis for judicial review of administrative decisions. See unfavourable decisions when they, nonetheless, perceive the procedure to be fair (an example could be a person who willingly accepts the unfavourable result of tossing a coin to decide upon an advantage in a sporting contest). The more unfavourable the decision, the more important is the perception of procedural fairness. 61 Conversely, the higher the degree of procedural fairness, the higher is the likelihood that the final decision will be willingly accepted. 62 Having proposed an understanding of what procedural justice entails, it is now time to analyse the OTP's selection procedure against three essential components of procedural justice: consistency, impartiality and representation. These three are said to reflect widely shared 'intuitions of justice'. 63 They emerge across a range of studies as the foremost indicators of procedural fairness. 64 In the words of Tom Tyler, 'people care about the decision-making process [and] they consider evidence about representation . . . bias [and] consistency'. 65 The present one is by no means an exhaustive analysis of procedural fairness, and no special weight is ascribed to the components against one another. For those in affected communities, these components are not, in themselves, likely to be decisive in the acceptance of a given selection. Indeed, at differing times, there may be a range of political, social and conflict factors at playnot to mention that some communities may well have traditional perceptions of justice that the ICC may not be able to satisfy. Ultimately, only an empirical assessment could provide community-specific answers about the Court's perceived legitimacy.
Nonetheless, the present analysis has predictive value and can inform the assessment of how affected communities' perceptions may take shape, and the role that procedure can play in this regard. Although affected communities are comprised of a diverse range of people, commentators suggest a tendency for groups -by way of the socialization of their beliefs -to understand procedural fairness in broadly similar ways. 66 Should this be true, the present analysis may help to establish a basis for the conduct of future empirical research across and within different affected communities.

A. Consistency
The idea of consistency denotes similar behaviour, performance or treatment, often over a period of time. 67 A consistent procedure demands the equal application of principles or rules to similar sets of facts. The concept of consistency is commonly traced to an Aristotelian principle of justice which proposes that 'like cases should be treated alike, and unlike cases should be treated unalike in proportion to their difference'. 68 This principle -'treating like cases alike' -has gained an axiomatic status, particularly in the context of nondiscrimination and equal treatment. 69 In legal settings, 'like cases' may be identified insofar as they share a certain description or display certain common features that can be determined by the applicable rules. 70 Nonetheless, there is considerable uncertainty as to how the maxim applies in practice, and that includes the degree of difference between cases that would justify different treatment. 71 In turning to the OTP, the present analysis suggests that affected communities are equally likely to see procedure treat like cases unalike rather than like cases alike.
The essential starting point for determining consistency is equality of treatment. In this context, treatment means following procedural steps in the course of selecting a case. The degree of consistency present then depends on adhering to the same steps before the selection of a case is finally made. Here, whilst 'like' cases should be subject to the same procedural steps, where cases are 'unalike', any differences in those procedural steps should be justified by reference to the proportion of differences between those cases. Of course, identifying sufficiently 'alike' cases is generally more onerous than identifying cases that are materially different. However, it is in this second regard that the current selection procedure lacks the requisite consistency, due to the absence of justifications for differential treatment. This is best evidenced from three features of the selection procedure: its duration; its deference towards national investigations and prosecutions; and its dependence on relative concepts like gravity.
First, there is considerable inconsistency between the various situations with respect to the duration of preliminary examinations -an initial but essential stage in the selection procedure. On the one hand, the disparities of time are to be expected given the context-specific complexities of the alleged crimes and the accompanying challenges of evidence-gathering and management of capacity/resources. 72 Nevertheless, consistency requires efforts to harmonize differences of treatment accompanied by convincing explanations for any need for differential treatment. Although the Rome Statute contains no specific provision that regulates the length of preliminary examinations, the Prosecutor has argued that the statutory silence was a deliberate choice on the part of the drafters to afford her Office flexibility. 73 However, as Pues discusses, the drafting history does not support such a conclusion and there would need to be compelling evidence to support what has transpired to be a limitless discretion regarding the duration of preliminary examinations. 74 A brief overview of the duration of a range of preliminary examination evidences the inconsistency. The preliminary examination(s) lasted: one week in the Situation in Libya; over two years in the Situation on the Registered Vessels of Comoros; approximately 10 years in the situation in Afghanistan and more than 13 years (and still ongoing) in the situation in Colombia. The Situation in Palestine offers a similar story of temporal inconsistency. First, the OTP took more than three years to determine that Palestine was not a state and therefore was not capable of accepting the Court's jurisdiction. It then took the OTP almost five years before, in December 2019, it determined there was a reasonable basis to proceed to an investigation. And yet, the duration of this examination is likely to be significantly prolonged by the OTP's recent request for a PTC ruling on territorial jurisdiction. 75 There may well be legitimate reasons for the time such examinations have taken, including the time necessary to identify potential cases for investigation. However, the OTP, whether in its yearly examination reports or in any public documentation, does not, and perhaps cannot, offer comparative explanations as to why each situation requires such differing times. Even former staff members have been at a loss to explain why the Colombian situation did not reach the investigation stage. 76 It is then hardly surprising that the OTP's request for a territorial ruling in the Palestinian situation was received sceptically and cast as an unnecessary delay, given that the question of territorial scope could have been litigated later, e.g. after an arrest warrant was issued or when proceedings had commenced. One might even argue that the OTP has opened the door availability of information, the nature, scale and frequency of the crimes, and the existence of national responses to alleged crimes. 73 The only reference to preliminary examinations is in Art for the Court to make a negative determination on territorial jurisdiction, limiting the prospect of an investigation being authorized. 77 Against this backdrop, the OTP's 'negative practice' in omitting to publicize reasons that justify differential treatment between situations entrenches the view that inconsistencies in time are a way to avoid politically contentious investigations, and prolonged delays are a way for certain cases never to be selected. 78 Secondly and relatedly, further evidence of inconsistency can be located in the differing degrees of deference afforded to national legal responses. The display of deference finds greatest expression in one of the components of the admissibility assessment -complementarity. 79 The OTP's selection procedure entails a subjective determination of whether the state is sufficiently unwilling or unable to 'genuinely' carry out an investigation or prosecution. 80 Crucially, the Office can evaluate the admissibility thresholds for however long it is necessary. 81 Such uncertainty of time is compounded by the OTP's policy of 'positive complementarity', by which the OTP actively endorses and promotes national criminal proceedings. 82 The policy is either understood as part of the Court's shadow effect that can lead to catalysing national proceedings (and promoting deterrence of future crimes) or it is understood as an indirect means of supporting domestic judicial capacity. 83 Either way, several aspects remain unclear: when such a policy is adopted among all current and potential 77  situations/cases, what the rationale is for pursuing the policy at any one time and, ultimately, how such a policy influences the exercise of discretion. The inconsistency in positive complementarity is best illustrated by a direct comparison between the OTP's respective interventions in the Situation in Kenya and the Situation in Colombia. The Office's policy of positive complementarity in the Situation in Kenya found expression in the early encouragement and patience it demonstrated towards the Kenyan authorities before, eventually, deadlines were imposed -missed -and then arrest warrants finally issued. 84 In the Situation in Colombia, the Prosecutor has sought to use positive complementarity as a tool to catalyse national prosecutions and otherwise monitor domestic proceedings akin to a watchdog. 85 Inconsistency is the result because the OTP has not convincingly justified its different approach to positive complementarity by pointing out differences between the situations. Furthermore, the policy's influence on the final selection decision is inherently uncertain when one considers former Prosecutor Luis Moreno-Ocampo's now notorious declaration that the mere absence of cases would demonstrate the Court's effectiveness, because it would imply national authorities were undertaking their own prosecutions 86 -an assumption that entirely overlooks the complexities of determining causality. In summary, the selection procedure gives rise to arbitrariness by the variation in deference afforded to the policy of positive complementarity.
Thirdly and finally, the greatest source of inconsistency in the selection procedure arguably lies in the determination of the gravity of situations and cases. The Rome Statute fails to provide criteria governing the exercise of discretion in selecting situations and so, in recognition of its limited resources, the OTP selects situations based on their relative gravity. 87 The OTP also uses relative gravity as a criterion for case selection, given that its objective is to focus on the most serious crimes of concern to the international community. 88 The OTP has declared that satisfying the threshold requires an assessment of both quantitative and qualitative criteria that relate to the scale, nature, manner of commission and impact of the crimes. 89  notion that is itself 'vague, nebulous and quintessentially subjective' 90 -is made more uncertain by its relative application to each set of facts. Considering that this relativity inherently invites one to identify the most marginal of differences to justify differing selections, then, it is of little surprise that the procedure is vulnerable to arbitrariness.
One way by which the procedure permits arbitrariness is by blurring the distinction between situations and cases. At the preliminary examination stage, gravity is examined against a backdrop of the likely set of potential cases that would arise from an investigation. 91 To illustrate, let us consider the OTP's decision not to proceed with an investigation in the Situation on the Registered Vessels of Comoros, Greece and Cambodia. The Office argued that said situation concerned a limited number of victims and drew a comparison with the Abu Garda case, which had a similar 'size'. The Prosecutor argued that Abu Garda was distinguishable because of its nature and its impact: it concerned attacks intentionally directed against African Union's peacekeepers, including the attempted killing of eight of them. Such attacks on peacekeepers would 'strike at the very heart of the international legal system established for the purpose of maintaining international peace and security'. 92 However, the persuasive criticism has been made that the OTP's approach conflated the distinction between the situation as a whole and potential cases. 93 Indeed, even the PTC disagreed with the OTP's gravity analysis in the Comoros situation. The PTC set forth the assumption that if events are unclear and conflicting accounts exist, then these factors militate in favour of sufficient gravity, and only a full investigation can determine the events that unfolded. 94 Put simply, there is little agreement on how gravity is to be assessed within the Court. 95 Even if one could defend the OTP's use of gravity to select cases, one is very unlikely to find reasonable explanations for differing treatment when the objective differences between situations/cases in terms of gravity are -for all intents and purposes -so contested and tenuous. To summarize, the OTP's selection procedure appears to lack consistency due to the absence of justifications put forward by the Office for the differential treatment given to various situations/potential cases. Consistency may not, by itself, be a prominent factor in affected communities' views of the Court. However, evidence of inconsistency can readily be pounced on and cast as something worse (e.g. bias), particularly when information trickles downvia the unsympathetic filters of political elites or media coverage -into affected communities. Leaving aside the fact that affected communities might not have -and might never have -sufficient knowledge of the OTP's selection procedure to assess its consistency, the fact remains that there is a gap in the Office's provision of justifications for differential treatment with regard to the duration of preliminary examinations, the degree of deference in its policy of positive complementarity and to its assessments of the gravity of situations and cases. This is not to imply that the OTP can merely fill this gap with extensive explanations -this would pose its own practical problems. Leaving aside the question of when and in what form such explanations could be given, the OTP would be likely to proceed with extreme caution in disclosing reasons that might be challenged in court. 96 However, so long as such lack of publicity continues, there is a continuing risk that affected communities may perceive that prosecutorial decisions are inconsistent rather than consistent.

B. Impartiality
From consistency flows the principle of impartiality: treating parties to or rivals in a conflict equally. 97 In most instances, impartiality refers to the 'state of mind' or virtue of a decision-maker who is overseeing a procedure (e.g. a hearing) between, typically, two parties. 98 Impartiality is a fundamental principle of justice because it reflects fairness and inspires public confidence in justice being seen to be done. 99 The term is distinguishable from 'neutrality', which describes the absence of any position in support of one party. 100 Therefore, being impartial does not necessarily imply neutrality, because the former permits taking a position provided that the parties receive equal treatment. 101 In addition, the concept of impartiality shares a crucial relationship 96  with the cognate one of independence. One can act independently but not necessarily act impartially. However, a general lack of independence will, invariably, provide grounds to question one's impartiality. 102 The antithesis of impartiality is the concept of bias: being unfairly prejudiced against particular individuals or groups, or unduly concentrating an interest towards an exclusive target or range of subjects. 103 Frequently, therefore, impartiality finds greatest expression in one of the principles of natural justice: the rule against bias. 104 An impartial procedure is one that demonstrates an absence of bias towards either relevant party. In this light, this section explains why affected communities may be more likely to see the procedure's treatment of parties as biased rather than impartial.
To begin, the present concern is that of apprehended or apparent bias, substantiated by ascertainable facts such as lack of independence, conflicts of interest or simply discernible behaviour or conduct -all of which result in a risk of actual bias. 105 Self-evidently, apprehended bias is based on the audience's perception. 106 In the words of the late Justice Scalia of the US Supreme Court, 'what matters is not the reality of bias or prejudice, but its appearance'. 107 For the purposes of this article, the question to contemplate is the risk of affected communities apprehending this type of bias as a result of the OTP's selection procedure.
To answer this question, one could use an objective or 'objectivized' test based on the perspective of a standard on-looker or observer. In the context of judicial disqualifications, many jurisdictions have developed hypothetical tests, for instance assessing whether 'the fair-minded and informed observer, having considered the facts, would conclude that there was a real possibility that the [decision-maker] was biased'. 108 An iteration of such a test was developed at the International Criminal Tribunals for the former Yugoslavia (ICTY) in the Furundžija case. 109 In common law jurisdictions (e.g. in England and Wales) and in civil jurisdictions (e.g. in Germany), these disqualification tests apply in reviewing the decision-making of public administrative bodies such as a prosecutor's decision to proceed with a prosecution or not. 110 The ICC's Appeals Chamber has adopted a similar test in relation to the Court's Prosecutor, assessing whether there could be an appearance of bias 'based on the perspective of a reasonable observer, properly informed'. 111 Such a construct is a heuristic device to determine whether, looking in from the outside, impartiality is outweighed by an appearance of bias. It is thus appropriate to ask whether there is an appearance of bias from the perspective of 'reasonable observers, properly informed' within affected communities. In adopting this test, one needs to ask two questions: (i) what is a reasonable observer? and (ii) what knowledge makes an observer 'properly informed'?
First, reasonable observers can be cast as those who would apprehend bias based on objective circumstances. 112 The observer's reasonableness would find expression in their fair-mindedness: a reasonable observer could be defined as someone who 'always reserves judgment on every point until she has seen and fully understood both sides of the argument', 113 'someone who is not unduly sensitive or suspicious' 114 or prone to making snap judgments or reaching hasty conclusions based on an isolated episode. 115 Secondly, informed observers have 'taken the trouble to inform themselves of all matters that are relevant within its overall social, political or geographical context . . . '. 116 However, they cannot be presumed to possess a detailed knowledge of the law beyond that acquired through ordinary life experience -though conversely they should not be imagined as being wholly uninformed about the law in general 117 and may be expected to be aware of the basics of legal traditions and culture. 118 When applying this criteria to the assessment of the ICC's selection procedure, reasonable observers in affected communities should be imagined as firmly putting their subjective preferences aside. 119 They should also be conceived as having been able to critically assess negative narratives about the Court expressed by politics and media. Moreover, the ICC Appeals Chamber described a reasonable observer as one who is 'properly informed, . . . aware of the functions of the Prosecutor'. 120 Being aware of the Prosecutor's functions, presumably, refers to not only having a basic knowledge of her dutiesinvestigating and prosecuting -but also of the way in which those duties are to be discharged. 121 A reasonable observer would be aware of the significance of independence for the Prosecutor's impartiality and of the expectation that the Prosecutor's behaviour is free from any bias or external influence. 122 However, if such a reasonable observer were to assess the Prosecutor's selection record to date, it is not unlikely that they could find an appearance of bias.
First, an observer could reasonably apprehend bias on grounds related to lack of independence and impartiality. At the institutional level, the OTP is embedded in the political realities of the Court's jurisdiction. 123 For instance, the OTP's freedom in decision-making is jeopardized because the UNSC can trigger the Court's jurisdiction. In doing so, the UNSC can, at a minimum, occupy the OTP's attention, shape its priorities and significantly influence the exercise of its discretion -and this despite the fact that the OTP, technically, retains the right not to proceed to an investigation. 124 At the individual level, the Prosecutor's own professional conduct has previously been brought into question, in particular that of former Prosecutor Luis-Moreno Ocampo. Concerns were raised with regard to the former Prosecutor's pursuit of the Court's first self-referrals by Uganda and the DRC, 125 interpreted to be an expression of a self-interested desire to put runs on the board by seeking cases that would be 'easy wins'. 126 The former Prosecutor's independence has also been reasonably questioned due to the public disclosure of his less than robust commitment to professional ethics. 127 These grounds, alone, are sufficient to locate an appearance of bias.
Secondly, there is a chance that an observer would still apprehend bias even if they heard the OTP's explanation of its selections. Preliminarily, one should note that even a reasonable and informed observer may be conceived as not possessing detailed legal knowledge, including an understanding of nebulous notions such as gravity. In this scenario, a reasonable observer would not be able to properly assess the OTP's explanations about its prosecutorial choices. All the more, an observer considered to be in possession of such elaborate legal knowledge would be unlikely to find the OTP's explanation of its prosecutorial choices as adequate. The Policy Paper on Case Selection and Prioritisation does little to explain the Office's selection procedure and simply confirms that the OTP retains considerable flexibility in its decision-making, 128 through practices such as case sequencing or prioritization. 129 The Policy Paper merely tends to justify past selection practice and, by contrast, sheds very little light on why particular cases -among those that could have been selected -were, in fact, chosen instead of those that were not. 130 In the context of several eligible situations/cases, all selections entail a choice of one over another alternative. However, the Policy Paper's position that all choices can be explained by objective criteria (e.g. the criteria under Article 53(1)(a-c) of the ICC Statute) either begs the question of why the hypothetical alternative was discarded or it completely (and unrealistically) fetters the OTP's discretion.
Of course, the fundamental problem is the very expectation that the Prosecutor's explanation about her choices should be considered at all. This Prosecution Selectivity and Procedural Justice 21 of 33 expectation effectively circumvents and misunderstands the problem of apparent bias. When more knowledge is attributed to those reasonable observers, the standard is made increasingly unrealistic. Put another way, the objective standard is rendered meaningless if one insists that observers should take into account knowledge they ought to have rather than the knowledge they would ordinarily be expected to have. 131 In so doing, the OTP -by seeking to continually explain away and deflect suspicions of bias by legalistic explanations -is simply 'holding up a mirror to oneself' 132 and exposes its inability to acknowledge or address reasonable apprehensions of bias held by those on the outside. All in all, it is almost inevitable that even reasonable observers, from specific affected communities, will at any one time apprehend bias. These apprehensions of bias may be traced to distributive concerns -i.e. the patterns of prosecution between differing states, regions or sides to a conflict -which is linked to an ideology of impartiality based on the assumption that different groups receive equal benefits and carry equal burdens. 133 Indeed, it may be an instinctively human response to measure bias in terms of whether an opposing group or side has been targeted. 134 Nonetheless, this is not the complete picture of impartiality. If it were, an explicit strategy of selection even-handedness (similar to the one adopted by Carla Del Ponte at the ad hoc tribunals) 135 might be the only route leading to a target audience's confidence in impartiality. Mégret persuasively argues that impartiality cannot be about 'dolling out blame to both sides': otherwise, it would lead to a 'stultifying and paralysing policy of not discontenting anyone'. 136 This section suggests that the OTP's selection procedure, as it has been shaped so far, may lead a reasonable observer to apprehend bias. How (if at all) such apprehension of bias might be meaningfully addressed by the OTP is a different question -one that is beyond the scope of this article.

C. Representation
The concept of representation designates acting or speaking on behalf of someone, or portraying someone or something in a particular way. 137 For Pitkin, the concept of representation is an act -i.e. re-presentation -and involves making someone, or something, present when 'not being present literally or fully in fact'. 138 In this regard, representation is understood as giving a voice to an absent constituency, and thus reflects a basic intuition of justice -the right to be heard. 139 In practice, this right requires a 'representative' to act effectively on behalf of the constituency. 140 In so doing, representation is built on a set of presumptions about the very capabilities of a representative or an institution, to act on behalf of a constituency so as to further their interests. 141 These presumptions require attention to be paid to formal representation; namely, ensuring that there are procedural arrangements permitting the representative to be genuinely responsive to constituency interests. 142 Furthermore, there needs to be consideration of the extent to which those arrangements permit, substantively, the input and participation of the constituency so as to ensure their interests are truly heard. 143 In this context, this concluding section explains why the OTP's selection procedure is not satisfactory in adequately representing the interests of affected communities.
First, the selection procedure lacks any formal arrangements ensuring the involvement of affected communities. The OTP has declared that it welcomes direct interaction with victims and victim associations at the earliest stages of its work, to help define the focus of investigations and develop an assessment of the gravity of the crimes (including their impact on victims and affected communities). 144 However, there are no formal rules governing this crucial interaction, and the OTP has acknowledged the need to develop and refine best practices to enable victims to make representations to the Office. 145 This deficit is brought into sharp focus when one considers that the Rome Statute variously seeks to consolidate victims as actors (rather than passive subjects) of international justice and confers a range of procedural rights, in particular that of victims' right to trial participation. 146 The OTP's selection procedure has no formal accommodation or participatory regime for affected communities to interact with the OTP more generally. This is not to argue that selection procedure should permit affected communities to be given a formal status en masse, akin to that of participating victims. Such an arrangement would be entirely unworkable, to name only one problem. However, one should acknowledge the reality that only a fraction of the total number of victims qualifies for formal participation. Thousands of victims will remain wholly unrepresented because of jurisdictional restrictions, technical ineligibilities, inability or unwillingness to apply for recognition of victim status, or simply bureaucratic hurdles established by the Court. 147 This procedural gap in fully representing the views of victims can be evidenced by one key, though under-used, stage in the procedure: the discretion to decline to investigate or prosecute in the 'interests of justice'.
The OTP has declared that reliance on the interests of justice provisions (i.e. Article 53(1)(c) and Article 53(2)(c)) is exceptional, and there is a general presumption in favour of an investigation or prosecution. Nonetheless, the provisions explicitly require 'the interests of victims', including the views of affected communities, to be given due consideration. 148 This is a considerable challenge for the OTP, which commits to a dialogue with victims themselves, local community representatives and other actors who can help determine the impact of investigations or prosecutions on those interests. 149 Furthermore, to understand the interests of victims comprehensively, the OTP seeks the views of respected intermediaries such as local leaders, civil society and international NGOs. 150 The impulse to listen to a broad church of persons and groups is, of course, positive. The act of representation requires all relevant voices to be heard but, unsurprisingly, a diversity of views is likely to emerge that may conflict and/or be based on sectional preferences. 151 In this context, the OTP's representation of affected communities is unsatisfying for two reasons.
First, the OTP's representation cannot reflect the complexity of affected communities' interests prior to a decision being made. On the one hand, this challenge can be traced to the basic premise of representation, that of 'speaking for others', and the resulting tendency to collate and homogenize voices, so as to make them easier to represent. 152 This finds expression in the frequent simplification of victims' interests as simply wanting convictions and punishment. 153 However, this collation overlooks the diversity of constituencies' experiences and excludes minority voices. At worst, it reduces these constituencies to passive objects. 154 This has been particularly evident when the Prosecutor has deployed the rhetoric of 'the victims' in an abstract and de-politicized fashion, to legitimate and justify its selections. 155 Though it is not going to be possible to fully represent the interests of all those who constitute affected communities, the impression that is all too easily left is that only specific 'convenient' voices are re-presented to fit a pre-determined agenda (i.e. a pre-determined decision). 156 Secondly and relatedly, the OTP's representation does not permit affected communities to exert an influence on selection procedure after a decision has been made. In public administrations, those affected by a decision are nearly always given the opportunity to make their case. This includes being consulted prior to a decision and being able to request a review after a decision, often 'with a view to procuring its modification'. 157 This finds expression in another component of procedural justice, namely 'correctability': using an appeal mechanism to review a particular decision. 158 Turning to the OTP, the decision to prosecute or to discontinue should, as far as possible, respect 'the concrete community that is the victim of the crime and that will have to live with the consequences of the decision'. 159 However, the OTP's ability to comply with such an indicator of procedural justice is limited, because there is no mechanism for all those affected to review a decision not to prosecute. 160 Such a state of affairs de facto excludes members of affected communities from entering a 'value-expressive' dialogue with the OTP, 161 i.e. a form of dialogue which would allow the voices of affected communities to be amplified, and their status correspondingly elevated. Engaging in such dialogue would crucially increase the likelihood of individuals being satisfied with the Court's final decision, even one that would later remain unchanged.
To illustrate the point, it is worth recalling that numerous forms of decisionmaking review mechanisms exist across national jurisdictions. 162 Under the UK's Victim's Right to Review Scheme, to take only one example, a victim can request a review of a decision not to prosecute before any recourse to a judicial review. 163 This mechanism includes a local resolution where another prosecutor will review the correctness of the decision and, by providing additional information and explanation, will either confirm or reverse the original decision. Furthermore the victim is entitled to request an independent review by a different body (an Appeals and Review Unit) that will look at whether the original decision was wrong, and whether a prosecution should be brought to maintain confidence in the criminal justice system. 164 Although not immune from practical challenges, the existence of such a mechanism strengthens an institution's commitment to victims' rights and improves its quality of representation, by building in an opportunity to learn, reflect and ultimately correct mistaken decisions. 165 Crucially, the mechanism ensures that victims develop their agency and commence a formal dialogue that allows them, directly or via their own representatives, to seek further explanations and thus hold decisionmakers to account. 166 It is not at all suggested that the OTP should, or even could, reproduce such a scheme in its selection procedure, but the absence of any form of review highlights a deficit in the affected communities' representation within the OTP's selection procedure.
This shortfall is even more striking because the PTC has proven to be unwilling to compensate for it. The PTC has, at least until recently, tended to offer a conservative and deferential review of prosecutorial decisions. 167 Indeed, those victims that can come before the Court do not enjoy automatic standing to trigger a review but can 'prompt the Chamber to consider exercising its proprio motu review powers with respect to a specific issue affecting the victims' personal interests'. 168 The PTC's record includes accepting the Prosecutor's assurances that no relevant decision under Article 53(1) or 53(2) (c) has been made, 169 or declaring its lack of competence to undertake a review of independent investigative functions -possibly for fear of encroaching upon the OTP's functional independence -e.g. when refusing to review whether the Prosecutor had taken appropriate measures to ensure the effective investigation and prosecution of crimes in Kenya. 170 In such cases, the Chamber has evasively declared that the 'appropriate addressee of victims' concerns . . . should be the Prosecutor'. 171 Conversely, when the PTC has adopted a more robust form of review, it has tended to marginalize the interests of victims.
There is no better example of this than the PTC's decision in the Situation in Afghanistan. Although now overturned by the Appeals Chamber, 172 the PTC had originally rejected the OTP's request to authorize a proprio motu investigation into alleged war crimes and crimes against humanity. 173 In a decision that was subsequently widely criticized, the PTC concluded that, in light of the lack of cooperation that the OTP had received, the chances of a successful investigation were so small that authorizing one would not serve the interests of justice under Article 53(1)(c). 174  be no more than aspirational and that an unsuccessful investigation would, in the end, create 'frustration and possibly hostility vis-à-vis the Court and therefore negatively impact its very ability to pursue credibly the objectives it was created to serve'. 175 Albeit the Appeals Chamber has now curtailed the PTC's power of review in such proprio motu matters, the course of proceedings in the Afghanistan situation notably exposed the challenge of achieving representation. To illustrate, much of the original criticism of the PTC's ruling came from those purporting to represent potentially millions of victims in affected communities in Afghanistan. 176 This led to several filings by victims in relation to the Prosecutor's appeal against the ruling. Rather confusingly, these filings were made by various representatives of different cohorts of victims, including by the Office of the Public Counsel for Victims, NGOs seeking to act as amici curiae and other groups of individual victims who assert they have the requisite standing to file a notice of appeal (citing various grounds) directly with the Appeals Chamber. 177 Much to the frustration of the victims, the OTP, in return, made submissions that no such right to appeal for victims exists, because victims are simply participants to the proceedings but should not be recognized as proper parties, and thus only have a right to be listened to. 178 The Appeals Chamber, in a preliminary ruling, found in favour of the OTP and declared that those victims had no standing to appeal the PTC's decision. 179 Judge Carranza, nonetheless, dissented and argued that the victims should be on 'an equal footing' with the Prosecutor in being able to appeal a decision that, effectively, foreclosed an investigation into alleged crimes perpetrated against them. 180 In support, she argued that the Rome Statute had to be interpreted in light of internationally recognized human rights standards, including access to justice and the right to an effective remedy; and she cited national laws, across common and civil law jurisdictions, that permit victims to challenge prosecutorial decisions that are deemed to be adverse.
Perhaps a prime example of the oft-cited tension between the OTP and victims, these proceedings equally depict a fundamental problem faced by the OTP (and the Court at large): how to adequately represent affected communities and the interests of victims that reside within them? Admittedly, one could argue that the representation of affected communities will -on one level -always fall short of being perceived as just. However, the Court itself has set a high bar in claiming that 'people most affected by the crimes should have the right to understand, to participate in, but also to have a sense of ownership of the justice process'. 181 The PTC ruling in the Situation in Palestine, which ordered the Registry to establish a system of public communication and outreach activities among affected communities and to establish a continuous system of interaction, is only one step in that direction. 182 Indeed, the ruling might be a symptom that, to date, the OTP's selection procedure has not created an adequate sense of ownership in affected communities. If anything, it has readily reduced them to spectators -a symbolic constituency, one that is simply the 'triggerer-off of the whole thing'. 183

Conclusion and Recommendations
Prosecution selectivity has been described as the 'greatest problem of international criminal justice'. 184 This article contributes to the literature by way of its focus on selection procedure from the perspective of affected communities. Vis-à-vis this target audience, the article critiqued the procedure's effectiveness against a measure of perceived legitimacy. Using a Rawlsian model of imperfect procedural justice, the preceding analysis explained the shortcomings of the ICC Prosecutor's selection procedure in being sufficiently consistent, impartial and representative. In turn, this lack of procedural fairness may reduce the likelihood that the OTP's selections are perceived as legitimate within affected communities. More broadly, this article argued that the OTP is unable to produce the 'fairest' possible prosecutorial decisions as to situations or cases -culminating in the conclusion that its selection procedure only makes a limited (if any) contribution to the Court's perceived legitimacy.
The article aims to trigger further reflection and research on the Court's ability to fulfil an expressive function, i.e. to convey a message that can help to educate and improve perceptions of international criminal justice in societies and among their communities. 185 Further scholarly attention is needed with respect to the messages stemming from the Court's practices and procedures long before any verdict is pronounced or any punishment is administered. 186 One way to categorize the present analysis would be as 'interpretive expressivism' -an analytical paradigm principally concerned with (i) how those practices and procedures can be aligned with the norms and values of relevant audiences and (ii) how the ensuing messages can ameliorate the Court's perceived legitimacy. 187 Adopting such a paradigm does not mean overlooking the very real challenge that ICC organs (including the OTP) face in deciding when to prioritize the needs of a particular audience. Nonetheless, critiques engendered by interpretive expressivism may help to unpack the Court's relationship with audiences and to explore its receptivity to target audiences' demands. Most of all, even if one were to be entirely sceptical about the Court's concrete potential to progress in this regard, it is the very process of 'looking out' and engaging with external audiences that provokes internal organizational improvements.
Returning, then, to more immediate matters, the practical question that follows is how to improve the OTP's selection procedure. It is not suggested that the selection procedure can be improved by way of marginalizing matterof-fact considerations such as evidence, capacity or resources, or to develop any solution that would encroach upon or fetter the exercise of prosecutorial discretion. The Prosecutor's discretion helps to protect her independence and its exercise is inevitably more art and judgment rather than scientific method. 188 It is likely that there will always be a myriad of complex but legitimate factors involved in making calls that, ultimately, are always caseand context-specific. 189 It is also not suggested that the OTP's procedure can be improved by setting pre-determined outcomes, i.e. by incorporating distributive justice principles as decision-making factors. 190 The implication of such an approach would be to explicitly prompt selections that are more geographically representative of alleged crimes across the world, and so counteract the perception that international criminal justice tracks the preferences of powerful states. 191 Of course, under Rawlsian 'imperfect justice' terms, such an outcome of greater selection parity may be desirable. However, to expressly shape procedure to that end would be to just chase an 'appearance of parity' and, thus, expose the Court to further charges of illegitimate motives or overtly discriminatory selectivity.
Instead, this article has made a case for the OTP to fully exhaust its commitment to procedural justice components. Accordingly, the most principled basis for procedural improvements should rest on the development of an organizational culture (including decision-making incentives) that tracks these public intuitions of justice. 192 In view of the above, the OTP should fully maximize the psychological effect of its selection procedure. This endeavour may be assisted by adopting the following recommendations.
First, the OTP should commit to more consistency in its treatment of situations, e.g. in the duration of preliminary examinations and its deference to a policy of positive complementarity. This does not mean the OTP should selfimpose a precise time limit for preliminary examinations or maintain a strictly uniform approach to positive complementarity. 193 Instead of blind uniformity, prosecutorial choices should be openly tailored and proportioned in a way that is justified by the degree of difference between situations. As discussed earlier, there are reasons to proceed cautiously, but a way forward could be the adoption of benchmarks or indicators by which the progress of all preliminary examinations can be readily compared, contrasted and ultimately judged. 194 A concise set of indicators would establish transparent standards that can harmonize the internal and external assessments of preliminary examinationsinformation that would then complement the qualitative yearly reports. By taking such steps the OTP would improve its consistency, especially as the demand for consistency will only increase as the Office undertakes even more examinations.
Secondly, the OTP could develop more consistency by avoiding an overreliance on gravity as the basis for its selections. The concept is highly elastic and, in the words of DeGuzman, 'simply does not have enough agreed content to provide convincing justifications for selection decisions'. 195 The OTP's application of relative gravity exacerbates its arbitrariness, and its use has already become tainted with politicized judgments that do not inspire confidence. In this regard, the OTP could anchor its steps towards consistency by articulating specific goals and priorities across and within its range of situations. This would require seeking consensus on those goals and priorities but, once established, it would help to guide later case selections. 196 Put another way, there is a need to acknowledge the following question: 'what does the OTP seek to achieve once its case selection decisions are added up together?' 197 Establishing situation-specific goals and priorities would provide a more transparent basis for the Office's explanation of its selection procedure's consistency.
Thirdly, the OTP should incorporate the 'reasonable observer, properly informed' test into its Policy on Case Selection. The test would provide a normative standard that would direct the OTP towards meaningful self-evaluation of its procedure and even enable it to 'check' the internal and often unconscious biases of its individual personnel. 198 The test would lend a personal and humanizing touch to its selection procedure and, although being only a legal construct, it would crucially help to 'bring the public into the room'. 199 By doing so, the OTP can use the test to express respect for the everyday opinions of outsiders like the public at large, 200 and acknowledge its own 'blindness. . . to the faults that outsiders can so easily see'. 201 The OTP should not, however, automatically discard a situation or case on the mere basis that the test would be satisfied (i.e. that a reasonable observer would apprehend bias). Instead, the point is to encourage the OTP to be more deliberative and less defensive in engaging with inevitable (and plausible) criticism of its impartiality, no matter what selections it ultimately makes. 202 In that respect, procedure matters. Adopting this test could trigger an improved external dialogue that properly acknowledges the perceptions of affected communities.
Finally, the OTP needs to engage in critical self-evaluation about a fundamental question: how can its selection procedure (best) represent the interests of affected communities? On the one hand, the OTP's relationship with those communities must respect the Office's independence: the Prosecutor does not act on behalf of communities in a manner akin to a defence lawyer acting on behalf of their client, including by following their instructions. On the other hand, the OTP must foster a sufficiently close relationship so as to adequately represent their interests. After all, it has committed to 'systematically address the interests of victims in the work of the Office, [seek] their views at an early stage and continue to assess their interests on an on-going basis'. 203 The difficulty faced by the OTP in accommodating a 'happy' medium between these