1.1. A blameworthy moral choice: the mental element in JCE III

Joint criminal enterprise (JCE) holds all those who agree to a common plan involving the perpetration of a crime responsible for other foreseeable offences that take place during the execution of the plan.Footnote ²⁰ I here use scholarly discussion of the so-called ‘third’ or extended variant of JCE to introduce fundamental principles about blameworthy moral choice in international criminal discourse, since so many scholars have openly deplored JCE III's tendency to ‘overpower the restraining force of the criminal law tradition’.Footnote ²¹ While these criticisms appropriately expose basic principles applicable to mental elements in modes of liability, their idealized vision of the criminal tradition's predominantly restraining character understates the sometimes major gaps between theory and practice in domestic systems and their effect on the development of unjustifiable international doctrine.

JCE has three strands. The first ‘basic’ form occurs where ‘co-defendants, acting pursuant to a common design, possess the same criminal purpose’.Footnote ²² An example would be a plan formulated by three soldiers to torture a detainee, where each of the soldiers carries out a different role (holding the victim down, preventing others from entering the room, and applying electrodes and controlling the current). This ‘basic’ form of JCE holds each of the soldiers responsible for the war crime of torture, even though the men guarding the door and restraining the victim do not satisfy the elements of the crime – like Eichmann, they do not personally perform the crime. The second ‘systematic’ form of joint criminal enterprise is a mere subset of the ‘basic’ form, and therefore adds little of great salience for present purposes,Footnote ²³ mostly because it also requires that the participants in the enterprise harbour the necessary intent to torture.

The third variant, however, descends into darker territory. Under JCE III, all participants in a joint criminal enterprise are responsible for crimes committed beyond those agreed, provided they are ‘a natural and foreseeable consequence of the common purpose’.Footnote ²⁴ Thus, the soldier manning the door is also convicted of torturing the victim, even if he believed he was guarding the entry to prevent enemy soldiers entering and only foresaw that one of his confederates might commit torture while they were in the premises.Footnote ²⁵ The great anomaly is not only that the lookout is punished for having perpetrated torture, even though he did not personally hurt a fly; it is also that he is convicted based on mere foresight – a standard well below that defined in the offence for which he is punished. The key point is that JCE III tolerates a sharp cleavage between the definition of crimes and modes of liability used to convict defendants of them – the two categories overlap, but not perfectly.

Whence did the doctrine come? True, the famed Tadić Appeal Judgement declared JCE III part of customary international law, but this finding was a mere reiteration of national principles. To the extent that the court purported to draw on custom, it cited cases convened by British, Canadian, and American military tribunals applying Control Council Law No. 10, as well as domestic courts within Italy, all of which originally applied national concepts of attribution.Footnote ²⁶ And, in any event, the ICTY explicitly affirmed that ‘international criminal rules on common purpose [i.e., JCE] are substantially rooted in, and to a large extent reflect, a position taken by many States of the world in their national legal systems’.Footnote ²⁷ Therefore, to the extent that international criminal courts and tribunals are applying controversial standards of attribution like JCE III, it is largely because they have imitated national equivalents.

The critics, however, have shown JCE III no mercy, largely on the grounds that the incongruity between the mental element for the mode and that required for the crime leads to a violation of the principle of culpability.Footnote ²⁸ Traditionally the offshoot of retributivism, culpability reflects a commitment to the idea that an individual's punishment must be calibrated to her personal desert.Footnote ²⁹ The immediate retort (that I heard many times from Anglo-American lawyers in practice) is that this focus on culpability is overly academic when national systems depart from the principle as a matter of course. If JCE III solves the Hitler-as-accomplice dilemma and furthers the noble aspirations of the international justice project,Footnote ³⁰ why should international courts moderate their use of the doctrine when so many major Western jurisdictions apply an identical concept?

In simple terms, the answer is that guilt matters. An individual cannot be instrumentally punished to pursue even noble policy goals. Although this notion dates at least to Kant, in the English-speaking tradition H. L. A. Hart famously reconciled it with utilitarian theories of punishment by pointing out a disparity between the objectives of the criminal system as a whole and the principles to be employed in attributing blame in concrete cases.Footnote ³¹ He illustrates the distinction with a striking example – even if your rationale for punishment within the system generally is deterrence, it is clearly morally vulgar to punish family members of those who carried out criminal offences, even if doing so has massive deterrent effects.Footnote ³² In a similar example of greater salience for international criminal justice, George Fletcher chillingly recalls that ‘[a]s the National Socialists well knew in controlling inmates in slave labour camps, occasionally hanging an innocent person effectively deters disobedience by other inmates’.Footnote ³³ Quite clearly, punishment without culpability is anathema to liberal notions of criminal law, even if it does promote deterrence or other desirable outcomes.

Therefore, culpability is central to any theoretically justifiable account of criminal responsibility, from retributivism to restorative justice.Footnote ³⁴ True, advocates of restorative criminal justice may calculate guilt slightly differently,Footnote ³⁵ but they are still committed to the notion that ‘only the guilty should be punished’.Footnote ³⁶ In fact, these philosophical commitments are so widely held that Mirjan Damaška plausibly claims that:

And yet, while this is true at the level of principle, it overlooks states’ sometimes prolific abdication from theoretical standards in practice and the genealogy of JCE in national law.

Unsurprisingly, international criminal courts mimic this schizophrenic relationship with culpability. When addressing the concept in abstract terms, they also adopt a formal rendition of the culpability principle, insisting that ‘the foundation of criminal responsibility is the principle of personal culpability: nobody may be held criminally responsible for acts or transactions in which he has not personally engaged or in some way participated’.Footnote ³⁸ The pledge is laudable but it also omits half the concept. An individual is culpable, not just because she participated in criminal acts or transactions, but also because she made a blameworthy moral choice to do so.Footnote ³⁹ So, already the tremendous incidence of strict-liability crimes within Anglo-American jurisdictions reveals a great distance between Damaška's understandable appeal to culpability, and the practice of states that habitually disregard it.Footnote ⁴⁰

How, then, is culpability to be measured? To begin, note that the content of requisite blameworthy choice varies from one international crime to the next. Indeed, the availability of different mental elements allows states, treaty-makers, and sometimes judges to define crimes in such a way that each prohibits distinct moral transgressions. For some crimes, recklessness or negligence will suffice, whereas others are markedly more demanding in order to signal the particular moral magnitude of the violation. In the context of genocide, for instance, the requisite choice is not simply to kill individuals; it also involves carrying out these acts with a corresponding intention to ‘destroy, in whole or in part, a racial, ethnic, or religious group’.Footnote ⁴¹ For many, this added psychological disposition is the quintessence of the crime – it is the element that distinguishes garden-variety murder from what Raphael Lemkin described as ‘barbarous practices reminiscent of the darkest pages of history’.Footnote ⁴²

One would think, then, that convicting someone of genocide without this special intent emasculates the crime. And yet, international courts have found that a member of a JCE could be found to have committed genocide, even though he merely foresaw that his colleagues might carry out the crime.Footnote ⁴³ For many scholars, this is theoretical heresy. David Nersessian, for example, describes JCE III as a form of ‘constructive liability’,Footnote ⁴⁴ a term he uses in contrast to direct forms of liability, because ‘theories of constructive liability . . . allow conviction for the same offense even though the requisite conduct and mental state are absent’.Footnote ⁴⁵ In articulating what makes these mechanisms objectionable, Nersessian draws on the principle of fair labelling.Footnote ⁴⁶ The concept was originally developed by Andrew Ashworth to describe the need for specificity in the label of a particular offence, rather than lumping together vastly different categories of offending.Footnote ⁴⁷ Even though this original purpose is less germane here, the underlying idea was that ‘[f]airness demands that offenders be labeled and punished in proportion to their wrongdoing’.Footnote ⁴⁸ Otherwise, an accused is stigmatized by preconceptions associated with an offence that do not match his personal responsibility.

While I doubt that fair labelling (in the sense in which critics of international criminal justice use it) deserves an existence separate from culpability, it does alert us to an important insight – the label of a crime is a key element of punishment that must match an accused's guilt, regardless of the number of years in prison an accused is to serve.Footnote ⁴⁹ This reading reinforces that conviction for a particular crime requires fidelity to its identity, which predictably is contained in the crime's definition. The physical and mental elements in the paradigm of the offence thus define what it means to be responsible for violating that prescription. Accordingly, convicting an individual of genocide for merely foreseeing the crime overpunishes. It misapplies the criminal label genocide, which is reserved for more blameworthy conduct, then misconveys a degree of responsibility that is not paired to the defendant's desert.

As Darryl Robinson convincingly argues, this is an aberration: ‘the [defendant] still faces the stigma of a conviction for committing genocide, while having satisfied neither the actus reus nor the hitherto indispensable mens rea for genocide’.Footnote ⁵⁰ The approach transgresses principles of culpability and fair labelling ‘by lumping together radically different levels of blameworthiness under one label’.Footnote ⁵¹ While I am less convinced that international influences explain this position (as distinct from a combination of the Hitler-as-accomplice dilemma and readily available domestic tools like JCE), we are likely to share deep misgivings about the national justifications for the doctrine – calculating culpability as ‘a package deal’Footnote ⁵² still escalates responsibility on policy grounds.

Jenny Martinez and Allison Marsten Danner also propose that ‘certain forms of joint criminal enterprise . . . that tolerate a reduced mens rea should not be used in cases involving specific-intent crimes such as genocide and persecution’.Footnote ⁵³ Here too there is a concern that the distinctive features of these serious crimes are ‘weakened by the lowering of the mental state to recklessness or negligence, as would occur in a Category Three JCE’.Footnote ⁵⁴ Although Martinez and Danner recommend closer attention to the principle of culpability in international criminal justice in order to bolster a fragile legitimacy, promote human rights, and achieve transitional-justice goals,Footnote ⁵⁵ their critique also underscores more deontological concerns for fairness to the accused – using JCE III to circumvent special intent inappropriately amplifies moral responsibility beyond the contours of the crime.

A final set of scholars reach the same conclusion, albeit on slightly different grounds. Antonio Cassese, for instance, argues that JCE III may not be employed in conjunction with special-intent crimes for two very compelling reasons.Footnote ⁵⁶ First, to do so would connote a ‘logical impossibility’,Footnote ⁵⁷ since one may not be held responsible for committing a crime that requires special intent unless one is proved to have the requisite special intent.Footnote ⁵⁸ Second, he convincingly argues that the ‘distance’ between the subjective dispositions of the primary and secondary offenders must not be dramatic if they are both to be convicted of the same offense, otherwise personal culpability ‘would be torn to shreds’.Footnote ⁵⁹ Thus, to preserve analytical consistency, all modes of liability must require subjective standards that are the same as those announced in the definition of each particular crime. Otherwise, modes of liability warp responsibility as distinct from merely attributing wrongdoing in line with the moral weight of the crime in question.

1.2. The fundamentals of action: failures to punish in superior responsibility

At the turn of the seventeenth century, the famed internationalist Hugo Grotius wrote, ‘we must accept the principle that he who knows of a crime, and is able and bound to prevent it but fails to do so, himself commits a crime’.Footnote ⁶⁰ The statement represented the beginnings of the modern doctrine of superior responsibility, but again it was not until domestic courts prosecuted the Japanese General Tomoyuki Yamashita after the Second World WarFootnote ⁶¹ that the doctrine began its meteoric rise to prominence within international criminal justice. Since then, the doctrine's modern popularity has stemmed, in large part, from its promise as a solution to the Hitler-as-accessory dilemma. To this end, superior responsibility must function as a mechanism through which the superior is deemed liable ‘for the crimes of his subordinates’.Footnote ⁶² Accordingly, superior responsibility emerged as a ‘mode of liability’ with domestic backing that promised to overcome the insurmountable deficiencies of complicity and accurately capture the true moral responsibility of the puppet masters in atrocity.

Once again, the origins of the concept were largely domestic. To conclude a meticulous study of Second World War jurisprudence governing superior responsibility (which, in turn, served as a foundation for modern iterations), Kevin Heller observes that:

Much has transpired since, but this history undermines a thesis that broad modes of liability are necessarily hatched internationally. On the contrary, international courts enthusiastically embrace far-reaching doctrine once prominent domestic systems grant them their imprimatur.

Confusingly, there are several definitions of superior responsibility within modern international criminal law but, crudely speaking, a subordinate's wrongdoing is attributed to the superior where she has effective control over the perpetrators of crimes and knew or had reason to know of their offences but failed to prevent or punish the perpetrators.Footnote ⁶⁴ Only a decade ago, however, Mirjan Damaška alerted us that an aspect of the old concept is inconsistent with first principles – convicting a superior of the same offence as his subordinate for merely failing to punish that subordinate violates culpability too,Footnote ⁶⁵ in this instance, because ‘the opprobrium attaches to [the superior] for heinous conduct to which he has in no way contributed’.Footnote ⁶⁶ Underlying Damaška's complaint lies the supposition that culpability presupposes personal participation in the wrongful conduct required for the crime of which the accused is convicted.Footnote ⁶⁷

Let us embark on a brief exegesis to explore the significance of this idea. Wrongful action requires that a defendant's actions ‘must reflect on him in a way that makes the kind of criticism communicated by the imposition of criminal responsibility appropriate’.Footnote ⁶⁸ Murder requires actions that lead to death, rape requires insertion of a penis in a vagina, theft demands appropriation, and so forth. There can be no mix and match – an appropriation cannot be a murder and a killing cannot be theft. Thus, it is not acceptable that someone convicted of a crime did something morally reprehensible, if that action has no bearing on the content of the offence with which she is charged. These principles reflect basic liberal aspirations – in order to guard against the prospect of thought-crimes, guilt by association, or punishment based on status, a wrongful act calibrated to the definition of the crime is widely regarded as ‘a primary candidate for a universal principle of criminal liability’.Footnote ⁶⁹

In traditional understandings, wrongful acts tend to divide into two camps. For one category of offences such as rape and fraud, conduct alone is sufficient, since the action itself constitutes the criminal harm.Footnote ⁷⁰ To extrapolate into the international sphere, the war crime declaring that no quarter will be given merely requires the order not to take prisoners – it matters not whether injured or surrendering enemy soldiers are subsequently massacred in Lawrence of Arabia style, since the announcement itself suffices to commit the crime. In the second category of crimes, however, proof of harmful consequences is required. And, for these harm-type crimes, ‘a causal connexion between some action of the accused and the specified harm must be shown in order to establish the existence of liability’.Footnote ⁷¹

Two contrasting theories question this traditional thinking from opposing extremes, both of which are useful for understanding international criminal law's philosophical stance on these issues. From one side, there are those who deny the conduct/harm division outright, arguing that causation is a quintessential element of responsibility across all criminal offences.Footnote ⁷² Rape is not restricted to the conduct of inserting one's penis into a woman's vagina without consent, but denotes ‘causing sexual penetration of the female’.Footnote ⁷³ By analogy, the war crime of declaring no quarter is actually causing bodily movements in the throat, mouth, and lips, which subsequently cause an announcement to one's troops that no prisoners will be taken in battle. On this account of the philosophy of action, one can never escape causal analyses, even for what are commonly known as conduct-type crimes. The distinction between conduct and harm therefore disintegrates, leaving causation as a universal ingredient in blame attribution.

Others reach the diametrically opposite conclusion by denying causation any legitimate role in determining responsibility. These arguments rely heavily on notions of moral luck – if we are committed to punishing people for what they deserve, surely they should not benefit from their luck.Footnote ⁷⁴ Why, after all, should a would-be murderer who shoots at her enemy be punished less, merely because the victim, by chance, dies of a heart attack seconds before the bullet hits?Footnote ⁷⁵ If we are serious about culpability as the metric upon which to judge responsibility, we must eliminate these types of fortuitous scenario from our calculus. To do this requires nothing short of abolishing harm as a touchstone for criminal responsibility and, as a result, eliminating causation from the criminal lexicon. In its place, criminal offences would always be inchoate in structure, making attempt the paradigm for criminal responsibility.

As a reflection of the inherent deference to domestic orthodoxy, international courts and tribunals reject both extremes. To illustrate, in determining the liability of leaders within the infamous Radio télévision libre des mille collines (RTLM) for instigating genocide, the Rwanda Tribunal distinguished instigation as a ‘mode of liability’ from direct and public incitement to commit genocide, which operates as a separate inchoate crime. Like the conduct-type war crime of declaring no quarter, the crime of direct and public incitement to commit genocide ‘is completed as soon as the discourse in question is uttered or published’.Footnote ⁷⁶ Whereas a showing of causation would be required to convict the radio owners for instigating genocide, for this inchoate crime a ‘causal relationship is not requisite to a finding of incitement’.Footnote ⁷⁷ So, in contrast to the theory that causation represents the hidden structure of all criminal responsibility, international courts side with domestic example.

Likewise, arguments from moral luck have no real currency internationally – harm indisputably matters in international criminal justice. All range of international crimes, from deportation as a war crime to extermination as a crime against humanity, are defined in ways that make the actual occurrence of harm necessary for guilt. For the former, civilians must be expelled across a border; for the latter, members of a civilian population must perish.Footnote ⁷⁸ In some instances, international courts explicitly reinforce the normative significance of harm by explicitly stating that international crimes are not inchoate and that liability is contingent upon proof that the intended harm materialized.Footnote ⁷⁹ In sum, international criminal justice is highly deferential to domestic tradition, both in its practice of distinguishing harm-type and conduct-type offences along traditional lines and in considering harm a theoretical centrepiece of criminal responsibility.

But why the mimicry? If we are truly committed to culpability as the guiding feature of blame attribution in international criminal justice, should we not have boldly dispensed with harm as the essence of international criminal responsibility, in favour of a system that better accounts for the problem of moral luck? Less ambitiously, is it not more coherent to dispense with the traditional distinction between harm-type and conduct-type offences, to acknowledge that causation is common to both? That international criminal justice adopts neither of these positions again reflects the pull of domestic influence, not some nefarious utilitarian agenda derived from its international political status. The absence of any domestic practice adopting these theories has obviated the need for international criminal jurisdictions to choose between competing models, leaving them to contentedly follow established domestic doctrine regardless of whether it accords with principle.

This leads us back to superior responsibility, where international criminal justice's traditionalism plays out most keenly. To begin, we must acknowledge that one might dispute whether superior responsibility is the ideal illustration of these philosophical principles in action in so far as it involves liability for an omission. There is broad dispute in criminal theory about whether omissions cause anything. On the one hand, there are those who consider that an omission is ‘nothing at all’,Footnote ⁸⁰ and that ‘nothing comes of nothing, and nothing ever could’.Footnote ⁸¹ Conversely, social theories of causation posit that we ordinarily explain omissions as having causal powerFootnote ⁸² – we have no problem, for instance, saying that ‘a lack of rain causes crops to fail’.Footnote ⁸³ The debate need not delay us here though, since both judicial and scholarly discussions of superior responsibility assume a causal structure and, perhaps more pertinently, the question is largely peripheral to our central focus on accessorial liability.

How, then, do these foundational principles play out within superior responsibility? Intriguingly, misgivings about causation have prompted international courts to offer two corrections to the law governing failures to punish. The first is largely cosmetic – in response to the complaint that failures to punish convict defendants of harm-type offences without establishing causation, international criminal courts began adding language to the pertinent sections of their judgments, professing that ‘an accused is not charged with the crimes of his subordinates but with his failure to carry out his duty as a superior to exercise control’.Footnote ⁸⁴ This language, however, has proved more of a smokescreen to ward off conceptual criticisms than a marked normative change,Footnote ⁸⁵ but the important point is that the rhetoric became important to quell theoretical unease with the doctrine's overreach and that causation was the grounds for the discomfort.

In contrast, the second judicial correction was more radical. Several Trial Chambers convicted military commanders of a separate offence of ‘failing to take the necessary and reasonable measures to punish’,Footnote ⁸⁶ then proceeded to hand down drastically reduced sentences commensurate with the commanders’ failure to act (as distinct from the harm associated with a crime such as torture, extermination, or genocide).Footnote ⁸⁷ This transformation of superior responsibility from a mode of liability to a separate lesser conduct-type crime was born of major theoretical misgivings – there is no logically plausible means of reconciling failures to punish with causation.Footnote ⁸⁸ How, after all, can a commander cause a crime that is already complete by the time she is impelled to act? The solution, then, was to dispense with causation by transforming failures to punish into a form closer to attempt.

For the same reasons, the vast majority of academics agree that failures to punish must constitute a separate conduct-type offence, given the impossibility of the commander's failure causing the subordinate's crime.Footnote ⁸⁹ With respect to a crime that subordinates have already committed, causality is ‘logically impossible’, since later events cannot cause earlier ones.Footnote ⁹⁰ And, as far as future crimes go, giving failures to punish a separate existence where crimes have yet to occur would require an absurdity: ‘to initiate prosecution of a crime that has not yet been committed.’Footnote ⁹¹ On this basis, the problem of causation within failures to punish ‘has not and arguably cannot be resolved’.Footnote ⁹² In sum total, using failure to punish as a vehicle for convicting the superior of the same offence as the subordinate is ‘largely disproportionate’.Footnote ⁹³

Admittedly, a minority of scholars in international criminal justice does reach the opposite view but, strikingly, their disagreement consistently attempts to reconcile failures to punish with causation rather than simply denying that the concept is necessary.Footnote ⁹⁴ Otto Triffterer, for example, argues that failures to punish are based on a double causal connection to the offence: the first flows from the superior's initial omission to control the subordinates; the second derives from the failure to exercise a ‘second chance’ to absolve himself by referring the matter to justice.Footnote ⁹⁵ Many find this explanation unsatisfying,Footnote ⁹⁶ but its attempt to justify the mode-of-liability view in causal terms surely highlights the significance that causation plays in any robust theory of blame attribution, at least for harm-type offences.

Only one scholar begs to differ. In her excellent article, Amy Sepinwall makes the strongest case for treating the failure to punish as a mode of liability by pointing out how ‘failure to punish can be read as an expression of his support for his troops’ act’.Footnote ⁹⁷ She argues that, because the superior intends the failure, he aligns himself with the subordinate's atrocity.Footnote ⁹⁸ In so doing, the commander compounds his subordinates’ offence, such that ‘he ought to be held criminally liable for it’.Footnote ⁹⁹ Although I have some sympathy for this explanation,Footnote ¹⁰⁰ it unjustifiably snubs the philosophical rationale that makes causation central to theories of criminal responsibility everywhere. Without it, we abandon the project of creating an objective connection between an accused's action and the harm to which international criminal justice assigns moral weight in calculating responsibility, leaving little principled protection against thought-crimes, guilt by association, punishment based on status, or other innovative doctrines that allow policy to supersede desert.

If international criminal justice is to become coherent, not harsh, causation is an indispensable element for the perpetration of all harm-type offences. There is, however, one final twist in this plot. For secondary parties, the harm/conduct distinction disappears because the derivative nature of the secondary party's liability creates a cause-like relationship.Footnote ¹⁰¹ The war crime of declaring no quarter is a conduct-type war crime (in so far as the consequences of the declaration are legally immaterial), but assessing whether a superior can be convicted of the crime for failing to punish a subordinate who made the announcement demands causation too. How else can we justify convicting the superior of this particular war crime, other than by showing that his actions made a difference to someone else's committing the offence? Thus, if the vast majority of scholars in international criminal justice assume valid foundations in their criticisms of superior responsibility, causality must be an element of all ‘modes of liability’ within the discipline.

2.1. The nature of complicity: two defining features

2.2. The mental element for complicity

The mental element for accessorial liability is highly debated and inconsistently applied at both international and domestic levels. Most international criminal tribunals formally insist that ‘the requisite mental element is knowledge that the acts performed by the aider and abettor assist the commission of the specific crime of the principal’Footnote ¹²⁹ and justify the adoption of this standard as the embodiment of customary international law. Conversely, the Statute of the International Criminal Court demands that the accessory assist the perpetrator ‘[f]or the purpose of facilitating the commission of such a crime’.Footnote ¹³⁰ The intensity of volition, then, formally distinguishes complicity within competing international incarnations of the doctrine. And yet, this formalism conceals the reality that neither of these standards is that actually applied most often in practice and, more importantly, that no single static standard is theoretically defensible.

To begin, note the competing rationale for the purpose and knowledge standards. The purpose standard is said to promote autonomy by precluding criminal impediments to otherwise lawful activities that depend on social interaction, especially business. The knowledge standard, on the other hand, promotes social control and the prevention of crime by demanding that agents take interventionist action when aware that their actions are enabling offending.Footnote ¹³¹ In its most ambitious guise, the knowledge standard posits that the potential aider ‘might be an educative or normalizing force that causes the would-be offender to change his mind’.Footnote ¹³² Given the magnitude of international crimes, the knowledge standard would seem the more reasonable rationale for our purposes, and yet both consequentialist accounts miss the deontological mark.

At the level of doctrine, the famed purpose/knowledge dichotomy glosses over a complex literature arguing that the accomplice liability actually involves ‘two dimensional fault’,Footnote ¹³³ which goes to the assistance provided, the perpetrator's intentions, and/or the ultimate crime facilitated.Footnote ¹³⁴ Later, I argue that it is possible to transcend these intermediary steps by inquiring whether the accomplice satisfies the mental element of the crime itself. For present purposes, though, it suffices to recognize that in international criminal justice, these types of technicality are completely overshadowed by the intensity of the contest between the knowledge and purpose strands. Thus, when the US Supreme Court was asked to hear allegations that a company was complicit in crimes perpetrated within apartheid South Africa, it faced a veritable deluge of argument for either side of the purpose/knowledge divide.Footnote ¹³⁵

The dichotomy, like many other international modes of liability, is inherited from international criminal law's domestic forebears. The reliance of the International Criminal Court (ICC) on purpose was (perhaps unfaithfully) drawn from a similar standard in the American Law Institute's Model Penal Code, which appears to require that an accomplice provide assistance with the purpose of facilitating the crime.Footnote ¹³⁶ In contrast, other international criminal courts have drawn inspiration from the vast array of national systems and earlier international precedents, which merely require that an accomplice provide assistance to the perpetrator knowing that her actions facilitate the crime. Therefore, to the extent that complicity is a doctrine deeply divided in international criminal justice, the schism mirrors an identical disharmony between domestic traditions.

Remarkably, both sides of the dichotomy are misguided. As a matter of pure doctrine, recklessness is the mental element for complicity most frequently applied by international criminal courts. This is evident, for instance, from the habitual inclusion within most international criminal judgments of the refrain that:

Clearly, awareness of a probability is constitutive of culpable risk-taking, not knowledge. It goes without saying that the two concepts are far from synonymous. To paraphrase Glanville Williams, becoming an accessory by providing assistance ‘knowing that a crime is afoot’ is quite different from helping ‘knowing that a crime may be afoot’.Footnote ¹³⁸ Could it be that the purpose/knowledge debate has ignored the true application of complicity in international criminal justice?

Apparently so. To cite but one representative illustration,Footnote ¹³⁹ one International Criminal Tribunal for the former Yugoslavia (ICTY) Trial Chamber first reiterated the widely cited proposition that complicity merely requires awareness that ‘one of a number of crimes would probably be committed’,Footnote ¹⁴⁰ then proceeded to convict a politician named Radoslav Brđanin of the war crime of wilful killing for issuing a decree that required the victims to disarm. The key mental element linking Brđanin to the wilful killing (of victims and by perpetrators he did not know in anything except broad abstractions) was that he was aware that the disarmament decree ‘could only be implemented by use of force and fear’.Footnote ¹⁴¹ But force and fear do not inevitably mean killing. Could his plan not be executed through beatings, torture, forced expulsion, and intimidation instead? The only analytically plausible explanation is that the killings were probable but not certain. This, as we suspected, is recklessness.

So, international criminal justice here tolerates a type of doublespeak, claiming knowledge but applying recklessness. This disparity between theory and reality is surely alarming, but we should not lose sight of its origins. As we have seen, a range of theories hypothesize that these types of disparity between rhetoric and practice arise from importing interpretative styles typical of human rights and law of war into international criminal law, the effect of moral outrage on interpretative technique, or the broader political aspirations associated with transitional justice that are said to drive hermeneutics in international criminal adjudication.Footnote ¹⁴² And yet, here, the doublespeak has a quite different genesis. Domestic criminal law in a range of countries that officially adopt the knowledge standard not only allows a strikingly similar application of the rule;Footnote ¹⁴³ some of their leading academics openly lament that this ‘introduces reckless knowledge as sufficient’.Footnote ¹⁴⁴ The origins, therefore, are entirely domestic.

Things only get worse on the conceptual level. On closer inspection, none of the three highly debated standards (purpose, knowledge, recklessness) is theoretically justifiable. Like other modes of liability in international criminal justice, all three violate the principle of culpability in certain circumstances because they all tolerate the imposition of a crime's stigma in situations in which the person convicted of the offence did not make the blameworthy choice necessary to be found guilty of that particular offence. Many point out the perversity of using JCE III to escalate blame for genocide in this manner, but what about instances in which complicity has an identical effect?Footnote ¹⁴⁵ With accessorial liability, individuals are also held responsible for genocide where they knew or were merely aware that genocide was one of a number of crimes that would probably be committed.Footnote ¹⁴⁶ These scenarios, which are actually more common in practice, violate culpability too. Tellingly, these violations are explicitly based on examples drawn from a host of Western systems.Footnote ¹⁴⁷

Why, then, are all these standards conceptually problematic? Let us consider the reckless standard of complicity first. On a positive note, recklessness is at least very candid about its function – it avoids complicity being rendered ‘a dead letter’.Footnote ¹⁴⁸ The argument hinges on the inability to know the future with certainty. Even if member X of a criminal gang provides her terrorist colleague Y with a nuclear warhead for a specific terrorist mission, X cannot know with certainty that a crime will transpire.Footnote ¹⁴⁹ Short of knowing that water will flow downhill or that the sun will rise in the east, no one can know with certainty what the future will involve, and surely few human actions ever acquire a degree of predictability anywhere close to the direction of the Earth's rotation or laws of physics.Footnote ¹⁵⁰ Thus, argue the recklessness advocates, we either deny that complicity exists for assistance in advance of the crime (because people can seldom know what others will do in the future) or we apply a standard closer to recklessness for accomplices who choose to undertake acts they know are inherently risky.Footnote ¹⁵¹

The competing criticisms of recklessness are, however, equally compelling. For many, embracing reckless complicity would require us to continuously vet those with whom we have dealings so that we can ensure that our interactions do not lead to potentially wide-ranging criminal harm. This, according to many scholars, would have the unsavoury consequence of creating ‘blank cheque responsibility’,Footnote ¹⁵² where the aider becomes responsible for all foreseeable consequences of their daily public interactions, transforming the average citizen into an ‘unpaid auxiliary policeman’.Footnote ¹⁵³ Beyond unduly infringing upon liberty and individual autonomy, a reckless standard of complicity would offend liberal notions of punishment and inhibit social intercourse.Footnote ¹⁵⁴ To return to Glanville Williams:

But is this always the case? Notice how neither side of this debate mentions desert, even though scholars of international criminal justice rightly hold the concept dear. To conform with desert and its analogue culpability, recklessness should be appropriate as a standard of liability for the accomplice when it is adequate for the perpetrator. After all, recall that we earlier measured culpability by referencing elements of each particular crime. The argument conveniently dovetails with claims that using recklessness as a standard for complicity where recklessness suffices for the crime in question would not imperil an individual's autonomy or chill normal social interchange any more than recklessness perpetration already does.Footnote ¹⁵⁶ Consequently, there is no generic difficulty in reckless complicity as such; it is really the application of recklessness across international crimes whose mental elements vary that is unduly harsh. At present, this is the dominant scenario in international criminal law.

The knowledge standard for complicity is just as objectionable, albeit for slightly different reasons. As we have seen, a primary objection to the knowledge standard is that it is an epistemological impossibility for the vast majority of human actions,Footnote ¹⁵⁷ which probably explains why international criminal jurisdictions follow the many national systems that allow knowledge to surreptitiously dilute into recklessness. All the same, the knowledge standard presents another special peculiarity – knowledge of what? The dilemma is that the offence itself does not help to answer the question. After all, many international crimes make no mention of knowledge whatsoever.Footnote ¹⁵⁸ So, as soon as one utters the word ‘knowledge’, a whole series of deeply complicated, highly debated, and ultimately inconsistent responses arise,Footnote ¹⁵⁹ born of:

Unfortunately, in certain contexts, this uncertainty has permeated into international criminal jurisdictions too.Footnote ¹⁶¹

Complication, however, is not the knowledge standard's worst fault. The greater concern is that knowledge also violates culpability since, in David and Goliath fashion, it too overpowers higher mental elements. As previously mentioned, knowledge usually suffices for conviction of international crimes requiring special intents,Footnote ¹⁶² allowing the weaker complicity standard to eviscerate the stronger character of the crime. Like genocide, other international offences from pillage to torture are said to demand a specific purpose, for which knowledge should be perfectly inadequate.Footnote ¹⁶³ For each of these crimes, the objective contribution to the crime must be carried out for a concrete purpose (to destroy, to extract information, or for personal or private use). Therefore, by merely requiring knowledge as a necessary mental state for conviction, the test harks back to the characteristics of other international modes of liability so many scholars denounce as excess.

Perhaps purpose is the solution for complicity then? Alas, the purpose standard only fares marginally better – it overcorrects, then veers from the path of culpability too. True, in many of the jurisdictions that seemingly embrace purpose, the overcompensation is intentional. The drafters of the US Model Penal Code, for instance, concluded that the purpose standard was the preferable mental element for accessorial liability in order to offset the indirect nature of the accomplice's contribution to the criminal harm.Footnote ¹⁶⁴ While this approach is often celebrated as a laudable liberal adjustment to normal principles, we should reflect momentarily on its implications – by this doctrine, an accomplice who acts in such a way that she not only satisfies the mental element of the crime but makes an essential contribution to its realization is absolved of liability. And yet, if we maintain principle, this too misapplies desert and miscommunicates responsibility.

For one reason, an accessory who assists a crime with intent, recklessness, or negligence is not responsible, even though these mental elements are by far and away the most prevalent within the criminal law. Why is our faith in culpability so easily shaken here? If the mental element set out in the criminal offence really does define the degree of culpability associated with a crime (as many excellent critiques of other modes of liability in international criminal justice assert), then should it not also define the accomplice's desert too?Footnote ¹⁶⁵ Otherwise, our method of identifying culpability is capable of manipulation based on arguments from policy, in ways we earlier rejected. If we adopt a more sound approach to culpability, then purpose is very frequently an unjustifiable overcompensation that, in the final analysis, leads to potentially serious underpunishment.

Coincidentally, a utilitarian concept of responsibility would support the same conclusion. For a utilitarian view of punishment, purpose is unattractive since deterrence is maximized (as therefore is crime prevention) by punishing those who are aware of even the slightest risk of harm.Footnote ¹⁶⁶ Admittedly, here, there are more complicated questions about over-deterrence, which require a careful calibration of complicity standards with the desire for free social intercourse, especially in the realm of business.Footnote ¹⁶⁷ Nonetheless, utilitarian concerns tend to militate against adoption of the highest conceivable notion of blameworthy moral choice (i.e., purpose) across the entire panoply of international crimes, since complicity can achieve greater deterrence for such tremendous harm by setting the mental element at levels much closer to that defined in a crime. If recklessness suffices for perpetration of an offence, demanding that the accomplice assist the crime with the purpose of bringing it about underdeters accomplices.

Moreover, empirical research suggests that, in many instances, members of the public believe that the accomplice is blameworthy even though they did not share the perpetrator's criminal purpose. The subjects of one survey reported ‘stark disagreement’ with the ‘elevation thesis’ (viz. the idea that the mental element in complicity should be elevated to purpose – that is, beyond that required within the paradigm of the crime itself).Footnote ¹⁶⁸ Instead, respondents assigned punishments to accomplices ‘who are knowing or even only reckless with respect to the criminal outcome in instances in which the elevation view would assign no liability’.Footnote ¹⁶⁹ We might doubt the extent to which these findings could be extrapolated across an international community,Footnote ¹⁷⁰ but the research does at least serve as further grounds for caution. Using purpose as the mental element for complicity may badly fail to match popular notions of responsibility, which would diminish international criminal justice's prospects of promoting reconciliation, transitional justice, or other desired objectives. From both retributive and utilitarian perspectives then, purpose fails.

Where does this leave us? If the preceding analysis is correct, all static standards of complicity are indefensible. All fixed mental elements for accessorial liability (i.e., purpose, knowledge, or recklessness) violate basic principles of blame attribution since in each there will occasionally be a marked departure from culpability when the elements of the crime do not match those of the mode of liability. In these instances, complicity distorts an accused's degree of responsibility, either by amplifying culpability relative to the elements of the crime with which she is held responsible or by artificially elevating culpability beyond its normal parameters to absolve her otherwise blameworthy conduct. In these situations, complicity will only conform with culpability out of chance couplings between mental elements within complicity and those required for crimes. Relative to desert, responsibility becomes arbitrary, replicating the very characteristics so many scholars deride in other international modes of liability.

The consequences are, needless to say, hard to overstate. First, these departures from fundamental principles of blame attribution are not nefarious creations of an illiberal international system; they are borrowed from domestic criminal systems that set bad examples. While JCE and superior responsibility's origins in domestic law are more easily concealed in only a portion of national jurisdictions, there can be little doubt that domestic criminal systems from the vast majority of the world adopt objectionable static mental elements for complicity. The oftentimes heated debate between proponents of these various standards has only obfuscated the reality that none is theoretically defensible, and that the solution lies in transcending rather than deepening the terms of the current debate. In this light, international criminal justice's major sin is not that it has allowed policy interests or interpretative styles from branches of international law to crowd out criminal standards, but more that it has showed its domestic antecedents too much reverence.

Thus, if international criminal justice is to acquire normative coherence, it must disassociate itself from objectionable domestic precedent. How would a defensible alternative look? Well, if the only defensible conception of accomplice liability is one in which the mental element is the same as that required for perpetration, complicity ceases to retain any independent identity over and above perpetration, at least at the level of moral choice. And, if complicity begins to dissolve into perpetration in this way, should modes of liability as a species not disappear along with it for exactly the same reasons? In other words, if any static conception of a mental element within a mode of liability violates culpability relative to the mental elements in crimes (which vary from one crime to the next), should we not abolish modes of liability altogether in favour of a more capacious notion of perpetration? To a large extent, the answer to this question depends upon how complicity fares with respect to the second fundamental element of blame attribution. But, as we will soon see, the answer is no more positive.

2.3. The physical element of complicity

We have established not only that complicity functions as a mode of liability in international criminal justice, but that many international crimes also require harm as a precondition for responsibility. Given these characteristics, causation is conceptually necessary to bind accomplices to proscribed criminality if there is any chance of placating the critics of other international modes of liability and, more basically, of respecting the principle of culpability. In other words, having attributed to complicity an ultra-orthodox status in international criminal law, we are left with a stark and seemingly intractable choice between only two options: we either accept that causation is an element of accessorial liability, in which case it shares common features with perpetration, or we conclude that complicity is acausal in structure, in which case it violates principles of culpability in ways that scholars of international criminal justice rightly find reprehensible.

Surprisingly, international courts opt for the latter of these approaches. The accepted position before international courts and tribunals is that ‘proof of a cause-effect relationship between the conduct of the aider and abettor and the commission of the crime, or proof that such conduct served as a condition precedent to the commission of the crime, is not required’.Footnote ¹⁷¹ Needless to say, this defies basic thinking in criminal theory. By this reasoning, international courts define complicity in such a way that it explicitly violates a principle that international criminal lawyers view as cardinal, a principle that they often employ to passionately censure the breadth of other modes of liability within the field, and a principle that theorists call foundational.

Thankfully, some action relative to the criminal harm is required of the accomplice in international criminal law. International courts and tribunals invariably stipulate that ‘the actus reus of aiding and abetting is that the support of the aider and abettor has a substantial effect upon the perpetration of the crime’.Footnote ¹⁷² But, alas, this merely adds new layers of ambiguity to already opaque waters. What could it possibly mean to have a substantial effect upon the perpetration of a crime without causing it?Footnote ¹⁷³ The puzzle is how courts can simply do away with causation in favour of this substantial-effect standard when, by all accepted wisdom, ‘there is no way of contributing to any result, directly or indirectly, except causally’.Footnote ¹⁷⁴

The origins of this substantial-effect test are obscure in international criminal law,Footnote ¹⁷⁵ but we can speculate. Perhaps the position was tacitly influenced by English criminal theory, which has traditionally harboured a marked distaste for the view that an accomplice could cause the perpetrator's crime. Following the seminal work of H. L. A. Hart and Tony Honoré on causation, the large majority of commentators within the English-speaking world have argued that the volitional actions required to convict the direct perpetrator preclude the claim that the accomplice too caused the harm.Footnote ¹⁷⁶ The perpetrator made a decision; this interrupts all earlier causal influence, and acts as an intervening cause. On this account, the accomplice's actions are no more the cause of a crime than the perpetrator's genes, family history, and socioeconomic background, all of which undoubtedly provide influence without overriding the perpetrator's blameworthy moral choice.Footnote ¹⁷⁷ Could this reasoning possibly explain the doctrinal ambiguity in the modern international criminal understanding of the relationships between causation and complicity?

It seems doubtful. While Hart and Honoré's work has proved seminal, a competing line of authority has long recognized that in some instances the perpetrator's actions join rather than break causal chains created by an accomplice.Footnote ¹⁷⁸ For example, if X pays a hit man to assassinate his wife, arranges for the wife to be at a specific location at a time he discloses to the hit man, provides the hit man with the weapon necessary for the crime, and subsequently disposes of his wife's dead body, there is little trouble in declaring that X caused his wife's death.Footnote ¹⁷⁹ Was X the actual perpetrator (in the sense of breaking the glass)? Obviously not; the crime was committed by the hit man. Nevertheless, we are by no means precluded from simultaneously holding the hit man responsible for the killing, given that we have little difficulty in saying that both X and the hit man caused the wife's death.Footnote ¹⁸⁰ So, if Hart and Honoré's vision of perpetrators always acting as intervening causes was the inspiration for the international rule, then the choice was poor.

Perhaps the motivation for abandoning causation in complicity stemmed from a different concern, namely that the crimes would have occurred whatever the accomplice did.Footnote ¹⁸¹ In many instances, complicity is over-determined in so far as the accomplice's assistance was readily substitutable for the assistance of someone waiting in the wings. To return to the Zyklon B example, if the Nazis had access to a long line of willing suppliers of chemical asphyxiants (or variants that had comparable effects), then it would be difficult to argue that the suppliers of Zyklon B really caused the unspeakable consequences their chemicals enabled. As one of the defendants claimed, had he not agreed to supply the chemicals to Auschwitz, ‘the S.S. would certainly have achieved their aims by other means’.Footnote ¹⁸² Which judge, who knew anything about the stunning efficiency of the Nazi regime, could doubt the claim? So if causation means ‘but-for’ causation then even if this firm did not furnish the SS with the means of exterminating humans, the horror of Auschwitz would still have unfolded almost identically. Thus, these particular vendors of Zyklon B did not really cause anything.

This position, however, presumes an ill-informed notion of causation. In the vast literature on the topic, over-determination features as a recurrent theme.Footnote ¹⁸³ Throughout this extensive treatment, over-determined causes are consistently treated as a form of causal contribution, not grounds for adopting a substantial-effect test in lieu of basic principles.Footnote ¹⁸⁴ In line with this reasoning, a slew of commentators from different legal traditions consider that sine qua non causation must be assessed relative to events ‘as they took place’,Footnote ¹⁸⁵ in order to avoid allowing defendants like these to wash their hands of responsibility. So, by selling vast quantities of chemical gases to the SS for use in Auschwitz, Dr Tesch and his colleagues made an important causal contribution to the mass killing as it actually transpired. After all, ignoring how things actually transpired would mean that no one could ever cause murder. Everyone eventually dies, so the serial killer merely modifies the time, place, and manner of an inevitability. Clearly, the modifications matter.

In fact, if there is a deeper unspoken influence in this perplexing international account of complicity, it may herald from an unlikely domestic source. In a surprising parallel with international principles, German courts apply what is described as a furtherance formula (Förderungsformel), according to which the aider and abettor need not have caused but must have actually furthered (tatsächlich gefördert) the perpetrator's crime.Footnote ¹⁸⁶ And yet, the vast majority of German academics strongly disagree with this approach on the predictable grounds that it unjustifiably discards causation.Footnote ¹⁸⁷ In fact, they are only consoled by the impression that the furtherance formula probably differs little in practice from causality, especially when causation is calculated based on ‘the harm in its concrete appearance’.Footnote ¹⁸⁸ The parallel is illuminating. Even if the influence of this German position on the international rule is unavoidably speculative, its incongruity with accepted theory should dampen our assurance that unprincipled international rules necessarily reveal the triumph of the international agenda over the restraining force of the criminal law – domestic departures from principle are ubiquitous.

If all this is sound, we are still left with the challenge of rescuing the international definition of complicity from the jaws of domestic incoherence. To achieve this, the standard international position requires inversion. If we remove the word ‘not’ from the accepted judicial reasoning,Footnote ¹⁸⁹ the legal position becomes that ‘proof of a cause–effect relationship between the conduct of the aider and abettor and the commission of the crime is required’, and having a substantial effect on a crime (in the sense of making a causal contribution to events as they transpired) is a form of causation. This quick (but admittedly major) fix protects complicity against the criticisms that other modes of liability have correctly endured within the field by ensuring that causality plays a role in allocating blame to the accomplice. But it still leaves one further matter conspicuously unexplored: why must the accomplice's effect be substantial?

At first blush, this requirement is just as bizarre as the others. On the prevailing account of causation, an action is either a cause of an event or it is not – why the extra element? Again, the pull of mainstream domestic notions of complicity probably explains the doctrinal position. In both Anglo-American and Continental traditions, concepts of proximity or normative attribution intervene to preclude responsibility where the causal contribution is trivial, remote, or unusual.Footnote ¹⁹⁰ A member of the public who opens the door of a bank to let in a robber,Footnote ¹⁹¹ a restaurateur who serves a murderer dinner prior to a killing, or an onlooker who encourages the beating of a man who subsequently dies in an accident on the way to the hospital all make causal contributions to criminal harm, but these contributions are deemed too remote to warrant criminal punishment. The substantial-effect doctrine precludes liability, even though the assistance in each of these scenarios unequivocally contributed to crimes as they transpired.

Perhaps the incorporation of the substantial-effect doctrine in international criminal law reveals the positive side of domestic influence, even if it is part and parcel of a dependence that sometimes has perverse consequences. In some instances, international criminal justice imitates bad domestic examples that transgress culpability; but at other times, domestic influences serve commendable liberal purposes. If normative attribution does explain the need for a substantial effect, then it might fit into the latter category, even if the process of absorption into the international is not more conscious than that which produces international standards of blame attribution scholars rightly reject. In either case, though, the international political agendas and interpretative cultures from other branches of international law appear to play only backseat roles to the driving force of preconceptions derived from domestic criminal law.

In any event, once we return to the substance of complicity, our analysis indicates that a defensible notion of complicity incorporates both causation and normative attribution (or its equivalent proximity). As soon as we recognize this, we are immediately drawn back into Gardner's ‘splendid paradox’: if these elements are common to perpetrator and accomplice alike, why are accomplices not simply a subset of perpetrators?Footnote ¹⁹² As Michael Moore asks, ‘[a]ll substantially cause the harm, so why is one treated as an accomplice and the others treated as principals?’.Footnote ¹⁹³ To answer this, we must next investigate whether there is anything that necessitates a distinction between perpetrators and accomplices at the stage of attribution and, assuming a negative answer, whether complicity (like modes of liability in international criminal justice generally) occasions more departures from coherent philosophical principles, more doctrinal uncertainties, and more hours of costly intellectual labour than it is worth.

3.1. An abstract theoretical defence

International criminal justice's response to the Hitler-as-accomplice dilemma has played out in three overlapping phases. Initially, superior responsibility emerged as the theoretical response, but its popularity was quickly surpassed by the rise of joint criminal enterprise as the new prosecutorial doctrine of choice. Even though both of these modes of liability were drawn from Anglo-American criminal traditions then incorporated into the corpus of international criminal law as ‘sui generis’ forms of responsibility in international law,Footnote ¹⁹⁵ they both overextend basic principles of criminal responsibility. As this became increasingly apparent, these initial solutions to the problem were denounced as illiberal, sending decision-makers, practitioners, and academics back to the drawing boards. Once again, they would look for domestic examples, this time drawing on the German doctrine used to separate perpetrators from accomplices as the next borrowed domestic solution. Arguably, this was another false step.

Early on in its existence, the ICC embraced the German doctrine of ‘control over the crime’ as a basis for differentiating perpetrators from accomplices.Footnote ¹⁹⁶ On this understanding, Hitler was a perpetrator because he had hegemonic control over the atrocities in concentration camps, leaving the camp guards, bureaucratic administrators, and vendors of Zyklon B as mere accomplices. Domestically, this theory of ‘control over the crime’ was viewed as a major advance on earlier objective and subjective notions of perpetration. The first of these viewed a perpetrator as someone who actually swung the machete, but this failed to account for the fact that a perpetrator could use an innocent agent to carry out a crime on her behalf. The second subjective theory focused uniquely on whether the actor takes the crime ‘to be his own’, but this calculation cannot be easily established, and allows a person who perpetrates the crime with their own hand to be described as an accomplice.Footnote ¹⁹⁷

This initial influence quickly led to a major invocation of German doctrine, even when certain doctrines were highly disputed domestically. The international experiment with ‘control over the crime’ was soon followed by the adoption of German theories of co-perpetration, hitherto rejected in international criminal justice. Likewise, indirect perpetration was spawned as a viable mechanism for accounting for the Hitler-as-accomplice dilemma, at least in certain circumstances in which the perpetrator's will was overcome by that of a mastermind – at headquarters, in a command post, or behind a desk far from the bloodletting. To cap off the unconditional embrace of German criminal theory in international criminal practice, the ICC even adopted a more controversial German notion of functional perpetration through a bureaucracy,Footnote ¹⁹⁸ even though one leading German theorist feared that this ‘may create more problems than it solves’.Footnote ¹⁹⁹

Could the same thing be said for modes of liability in toto? Aside from the wider concern that such uncritical domestic transplants replicate the failed methodologies of the first two solutions to the Hitler-as-accessory dilemma, it also conceals major philosophical assumptions. One is especially important. Is there is any necessary distinction between perpetrators and accomplices? Perplexingly, the ICC treats the question as axiomatic – as if it is beyond all dispute. For instance, the decision that first adopts the doctrine into international criminal justice simply states that ‘the definitional criterion of the concept of co-perpetration is linked to the distinguishing criterion between principals and accessories to a crime where a criminal offence is committed by a plurality of persons’.Footnote ²⁰⁰ We are told nothing more about why the distinguishing criterion is so conceptually inevitable, perhaps revealing a blind spot in the enthusiasm for law crafted in a particular domestic system.

In reality, the division is far from conceptually inevitable. A number of other jurisdictions happily dispense with it (and its attendant technicalities) in favour of only the two elements used to dissect international modes of liability like JCE and superior responsibility. What matters on a unitary account of perpetration is that the assailant made a substantial causal contribution to a prohibited harm while harbouring the mental element necessary to make him responsible for that crime. To be clear, I do not consider that the unitary theory of perpetration is the only defensible account of perpetration or that differentiated models are inherently harsh. Quite the contrary – my ambition here is simply to point out that a unitary theory is at least as conceptually coherent as its counterpart.

To start, notice that the real question is not whether there is a moral difference between perpetration and complicity but if there is such a difference, whether it must feature at the initial stage of determining liability rather than later during the sentencing phase. There is no obvious structural impediment to taking accessories’ generally relatively lesser culpability into account at the sentencing phase along with other factors that are important to culpability but extraneous to the label visited upon the accused. If a defendant's motive for the crime, co-operation during trial, or history of recidivism are factors that appropriately reflect on the sentence they deserve, what rationale exists for treating the defendant's ‘important criminal energy’ differently?Footnote ²⁰¹ Certainly, it is difficult to see why one would choose to maintain an overly complicated, ever-expanding, and occasionally harsh set of ‘modes of liability’ in international criminal justice if a more streamlined system can also mitigate punishment as necessary.

To a large extent, this approach answers those who view perpetration and complicity as inherently distinct. For John Gardner, for instance, there is something innately privileged about being a perpetrator as compared with ‘mere’ complicity,Footnote ²⁰² such that ‘the attempt to eliminate complicity from the moral landscape, in favor of a more capacious domain of principalship, fails’.Footnote ²⁰³ The distinction between principals and accessories, he argues, is embedded in the structure of rational agency – there is a moral split between what one must do simpliciter and what one must do by way of contribution to what someone else does. While I tend to doubt the veracity of that claim,Footnote ²⁰⁴ observe how it does no work to maintain the segregated notion of complicity he defends – if there is such a distinction, it could figure at sentencing once the crime in question is coherently settled. Thus, the metaphysical distinction may well exist, but it has no obvious relevance for or against the unitary theory of perpetration.

Similarly, the derivative nature of complicity is also neutral as between differentiated and unitary models of perpetration.Footnote ²⁰⁵ George Fletcher, for instance, insists that ‘[p]erpetrators or principals are those who are directly liable for the violation of a norm; accessories are those who are derivatively liable’.Footnote ²⁰⁶ This is perfectly unobjectionable as far as definitions of a differentiated system go, but the definition does not purport to address (let alone justify) the partition of forms of attribution into a differentiated model. Clearly, we cannot escape our earlier analysis of derivative liability entirely – an individual's responsibility for the glass others broke will always be at least partially derivative of at least one other person's wrongdoing,Footnote ²⁰⁷ but nothing impedes treating the considerable assistance Eichmann or others provide to direct perpetrators as one of very many means of perpetrating an international crime. Perpetrators can also be all those who contribute to a crime, whether directly or through another.

This brings us to grammatical arguments. For many, the literal construction of certain offences uses terms that only a certain class of perpetrator can satisfy, creating a category of crimes often dubbed ‘non-proxyable’.Footnote ²⁰⁸ The classical illustration of non-proxyable crimes include bigamy (which only married people can perpetrate) and being drunk and disorderly in a public place (which only drunk people can perpetrate), so the argument is that a sober or unmarried person who assists these offences cannot perpetrate the crimes. If these examples seem too distant from international criminal justice, consider Pauline Nyiramasuhuko, the Rwandan minister of women's development, who was found guilty of rape for ordering militia under her influence to sexually violate Tutsi women by the thousands.Footnote ²⁰⁹ For many proponents of the ‘non-proxyable’ problem, convicting Nyiramasuhuko as a perpetrator of rape intolerably pretends that she has a capacity that she does not have – if rape is defined as requiring the insertion of a penis into a woman's vagina, she cannot be a perpetrator.

Yet, this reasoning holds little normative value. This is because complicity replicates the problem it is employed to solve – Nyiramasuhuko is found guilty of rape as an accomplice too.Footnote ²¹⁰ So, if accessorial liability also fails to solve the non-proxyable problem, crimes of these sorts are of no value in delimiting principal from accessorial liability. True, recasting complicity as the inchoate crime of criminal facilitation might solve the problem outright (because Nyiramasuhuko would not be labelled a rapist) but, as I have highlighted earlier, this ignores the sometimes tremendous harm that accomplices’ actions actually facilitate and discounts the fact that modern international criminal law unfalteringly treats complicity as a means of participating in the perpetrator's crime, not as a separate inchoate offence. As such, we can only soothe our anxieties about the non-proxyable problem by understanding that it is inevitable in a system that views harm as morally significant.

Indeed, the problem remains unresolved in ordering, instigating, JCE, superior responsibility, indirect perpetration, control over the act, co-perpetration, and functional perpetration – in short, all modes of liability. With each of these doctrines, the perpetrator need not satisfy the physical element in a criminal code. Why give Nyiramasuhuko's physiological status special significance over other physical elements in the definition of the crime? Doing so assumes an objective theory of perpetration: the assumption that the perpetrator is only the person who pulls the trigger, releases the gas, or, to borrow Eichmann's metaphor, breaks the glass. And yet, the objective theory of perpetration is entirely discredited elsewhere,Footnote ²¹¹ and the struggle for defensible solutions to the Hitler-as-accomplice dilemma in international criminal justice is a testament to its inadequacy in practice.

This, admittedly, involves tolerating a type of fiction (here, that Nyiramasuhuko inserted a penis into a woman's vagina, when she did not). This fiction could, of course, be quickly overcome by redefining rape (and other international crimes) in causal terms (i.e., as ‘causing rape’)Footnote ²¹² but even absent this kind of major legislative exercise, ‘what matters morally is significant causal contribution, not the kinds of limitations marked by the causative verbs of English’.Footnote ²¹³ To absolve Nyiramasuhuko of liability for the mass rape she caused based on a reference to physical attributes in the offence that she does not possess is to prefer fidelity to verbal semantics over substantive coherence. Like others, I believe that ‘normative rather than linguistic considerations would seem the more persuasive’.Footnote ²¹⁴ The overarching point, however, is that whatever real problems non-proxyable crimes present are equally true of unitary and differentiated models of perpetration alike, depriving this line of reasoning of any analytical purchase in debates between the two models.

Another frequent argument is that the unitary theory of perpetration violates the principle of legality by conferring judges with undue discretion in sentencing. For many, if statutory offences were meant to include even remote causal contributions, the legislature would have to enact even greater sentencing ranges below usual minimums, conferring judges with extreme discretion in sentencing.Footnote ²¹⁵ This renders law insufficiently certain. Worse, the breadth of this discretion would be all the more worrisome, as judges would still have to make the types of intricate differentiation that are presently undertaken at the attribution level (i.e., distinguishing between aiding, co-perpetration, indirect perpetration, instigation, etc.) when calculating sentences, only these determinations would take place without conceptual guidelines and behind closed doors.Footnote ²¹⁶ Consequently, a unitary system just brushes the problems under the carpet.

But these arguments are also unpersuasive. Although the point is obscured by the ICC's assumption of the differentiated model, the truth is that in the many differentiated jurisdictions, judges reason inductively to force facts into legal categories that they feel allow for an appropriate punishment.Footnote ²¹⁷ Instigation is elevated to indirect perpetration; aiding is recast as co-perpetration. To illustrate, certain courts describe a case of acting as a lookout for a criminal perpetrator – everywhere the textbook example of aiding – as co-perpetration in order to allow for the full scope of punishment afforded a perpetrator.Footnote ²¹⁸ In fact, this trend is so dominant in practice that ‘[i]t seems that no longer the dogmatic categorization determines the severity of the sentence imposed, but conversely that the severity of the sentence deemed desirable determines the categorization of the conduct in question’.Footnote ²¹⁹ If this is true, fears of judicial discretion are unavoidable for both theories, meaning that the argument from legal certainty does not lead inexorably to a differentiated system of perpetration.

Finally, arguments about the expressive capacity of a differentiated model do not appear to furnish it with great legitimacy. Under a differentiated scheme, a defendant's responsibility is expressed through the combination of at least two essential components: (i) the mode of participation and (ii) the name of the crime with which she is convicted. To eliminate (i) returns us to the problem of the fair-labelling principle – the criticism is that grouping the people who shot helpless refugees with AK-47s and the businessman who supplied the weapons under a single banner of, say, murder unfairly groups disparate degrees of responsibility, which a fair system of representation ought to segregate by employing additional qualifiers.Footnote ²²⁰ This observation leads into often implicit normative differences between perpetration and participation,Footnote ²²¹ and provides the impetus for the invention of notions like functional perpetration that allow the doctrinal label to encapsulate an element of the collectivity through which the crime came about.Footnote ²²²

There are, in my view, at least four problems with this account, each of which shows how a unitary theory is arguably more capable of fine-tuned expression than its counterpart. First, and least importantly, the argument for the expressive capabilities of a differentiated model ignores that international courts and tribunals do not mention the mode of liability within the disposition of their judgments in more that 95 per cent of cases surveyed.Footnote ²²³ As we have seen, in the vast majority of instances, dispositions contained in international judgments merely list element (ii), the crime with which an accused is convicted.Footnote ²²⁴ That there is almost never mention of element (i), viz. the mode of liability, within dispositions tends to suggest an unfortunate mismatch between abstract theorizing and practice – international criminal justice does not presently offer a vehicle for the expressive capacity that the differentiated model promises.

Other challenges are more difficult to overcome. Take, for instance, the miscommunication of responsibility inherent in labelling an accomplice a génocidaire when she does not have the requisite special intent. Some justify this disparity by pointing to a normative divide between ‘committing’ a crime and other forms of ‘participation’,Footnote ²²⁵ although the rationale for the division invariably goes unannounced. In purely analytical terms, though, one struggles to see a justification for the division when commission and participation both make an accused responsible for one and the same crime. As a matter of logic, the amalgamation of misaligned modes of liability and elements of crimes must corrupt one or both concepts, and branding an accomplice with a label she does not deserve still misrepresents responsibility, even if you have diminished the time she will serve in prison.

A differentiated model uses legal terms to express graduated degrees of blame, but there is also a danger that labels for modes of liability need not carry any great meaning for relevant audiences, further undermining the differentiated model's expressive capacity. Arguably, describing someone as an ‘instigator’ of genocide means something comprehensible to lay stakeholders in certain jurisdictions, but I have grave doubts about whether murder through a combination of indirect perpetration and co-perpetration holds any comparable significance. The risk is that the meaning of increasingly abstract legal terms used to describe modes of liability seems esoteric to ordinary citizens, who no longer understand the terminology or its moral implications.Footnote ²²⁶ So, when the ICC indicts former President Laurent Gbagbo as an indirect co-perpetrator, the BBC places the mode of liability in parentheses to mark the technocratic reference to legalese that only a small élite (who has no personal stake in the injustice) is likely to understand. If this is true, a lack of comprehension among the public adds another layer of distortion to the condemnatory aspirations of international trials.

Contrary to usual expectations, the unitary theory may offer greater expressive capacity. Under the unitary model, an accused could be convicted of genocide, denoting that she made a substantial contribution to the destruction of an ethnic group with the requisite intention to bring the crime about; then, a judgment could append a single concise plain-language explanation of her contribution, namely GUILTY of genocide for supplying machetes to the Interahamwe. Structurally, one would immediately know that this conduct led to the crime described and that the defendant adopted a subjective disposition necessary to constitute genocide – she really wanted the Tutsi exterminated. The differentiated alternative (i.e. GUILTY of (i) aiding and abetting (ii) genocide) does not tell us nearly as much about the culpability of the accused, because the formalistic concept ‘aiding and abetting’ varies so widely from jurisdiction to jurisdiction and, just as importantly, may spoil the identity of the crime.

This leads to a final related observation. To date, much of the debate around modes of liability in international criminal justice seems to have presumed that the crime and the mode of liability must do all the expressive work. But what prevents the judgment itself shouldering some of this load? For example, functional perpetration is necessary, we are told, to symbolically denounce the collective apparatus that enabled the individual crime.Footnote ²²⁷ And yet, this begs the question of why a court could not simply state whatever collective structures enabled the offence as part of its narrative. Without addressing this question, these often very insightful analyses of how traditional notions of perpetration do not adequately capture the reality of collective action that are so frequently part and parcel of atrocity risk overburdening ‘modes of liability’ when a plain-language explanation within a judgment may suffice.

3.2. The specificities of international crimes

If a unitary theory of perpetration is not theoretically foreclosed, we must inquire which of the two models is preferable for the particularities of international crimes. From the very beginning, the fact that an international system of blame attribution does not already exist is surely anomalous – for all the international interest in ending impunity, in transitional justice, and in modes of liability, there is no treaty regime that defines modes of participation in international crimes. With war crimes, for instance, the Geneva Conventions themselves furnish ‘only keywords to designate a criminal act, nothing which can be called a definition’,Footnote ²²⁸ leaving a range of indispensable criminal concepts ‘under a cloud of obscurity’.Footnote ²²⁹ This is most certainly true of modes of participating in these crimes – while the Conventions require states to implement legislation allowing for the prosecution of those responsible for ‘committing or ordering to be committed’,Footnote ²³⁰ they deliberately stop short of elaborating on the extent of these concepts.Footnote ²³¹ Whatever might be said about the merit of this approach as a means of securing broad participation in the treaty regime, it has proved to be a thorn in the side of practitioners ever since.

Despite popular views to the contrary, the ICC Statute does not markedly change this situation. For one reason, some of the world's leading countries are not party to the ICC Statute, meaning that recourse to customary international law remains inevitable in many instances in which international crimes might be enforced. As a reflection of this, the ICC Statute formally safeguards the continued co-existence of customary international law outside the treaty regime,Footnote ²³² allowing for a complex mosaic of blame standards that stem from all range of international legal sources to simultaneously cohabitate the discipline. Even states that have signed and ratified the Rome Treaty are not required to emulate modes of attribution as defined in the ICC statute within their domestic legal orders. As a consequence, international modes of liability are extremely difficult to identify.

The first problem with the scheme is methodological. While the ICC Statute brings a degree of clarity to cases arising within its four walls, many international trials still depend on custom as a source of law. The difficulty is that, as Martii Koskenniemi famously argued, custom is quite ‘useless’ at generating definitive standards.Footnote ²³³ So, even in a field like international human rights, where legal precision is comparatively less important, some of the leading exponents observe that ‘the human rights movement's quest for additional sources finds its favorite candidate, customary international law, in the midst of a profound identity crisis’.Footnote ²³⁴ Despite this crisis, a differentiated system of blame attribution depends on the very same mechanism for defining the terms of serious criminal responsibility.

This reality leads to major practical difficulties. For instance, despite the prolific use of JCE III over the past decade, the Extraordinary Chambers in Cambodia recently disagreed with the original Tadić decision that had declared JCE III part of customary international law.Footnote ²³⁵ Whatever might be said about the relative strengths of either court's reasoning, the content of customary modes of attribution is clearly unacceptably uncertain if different judicial bodies can reach diametrically opposed conclusions based on similar materials. While it is arguably not the business of international criminal justice to overcome the latent deficiencies with customary international law writ large, the absence of any restriction on the number of ‘modes of liability’ enables these types of scenario to continue unchecked. Put differently, a unitary theory of perpetration precludes the uncertainties of custom infiltrating the criminal process.

The wider concern is that such an ill-defined set of differentiated ‘modes of liability’ violate the principle of legality. As we well know, the principle of legality has a rich but troubled history in international criminal law, from its identification as merely a principle of justice at Nuremberg to more definite modern accounts that sometimes do not restrain any better.Footnote ²³⁶ When the methodology for identifying customary standards is so vague, and there is no numerical cap on how many modes of liability might be ‘discovered’ in customary international law, the danger is that international blame attribution seriously threatens legality.Footnote ²³⁷ As Beth Van Schaack points out, most common-law jurisdictions prohibited the notion of common-law crimes in the nineteenth century, precisely because the combination of judge-made law and serious criminal liability was perceived as compromising legality and its liberal underpinnings.Footnote ²³⁸ And yet, modern international criminal justice not only permits the historical anachronism; it also places no limit on the number of modes of liability the flawed methodology can generate.

Even where the principle of legality is honoured, standards remain seriously fragmented. Complicity itself is an illustration. If we accept the differentiated system incorporated in the ICC together with the German mechanisms for dividing perpetrators and accomplices, we are still left with a mental standard for complicity in the ICC that is markedly higher than the equivalent in the vast majority of crimes within that court's jurisdiction, with standards before other international courts that claim knowledge but contract to recklessness in practice (thus violating culpability in certain circumstances) and with all range of domestic variants of complicity across the spectrum of national courts capable of trying international crimes. If we are serious about expressive accounts of international justice,Footnote ²³⁹ a real danger is that the meaning of international condemnation is lost in translation.

We are then drawn back into the earlier discussion about the superior expressive capabilities of a unitary theory. In our previous theoretical discussion, we observed how a unitary theory allows a principled determination of criminal responsibility, and flexible opportunities to describe the nature of the contribution without legalese. If the expressive value of differentiated modes of liability is not fully comprehended within the national jurisdictions in which they originated, then they are likely to export very poorly to foreign cultures as part of the international adjudicatory process, given that victims, perpetrators, and members of their communities are even less familiar with the significance of terms like ‘instigation’, ‘joint criminal enterprise’, or ‘indirect perpetration’.Footnote ²⁴⁰ In the words of Immi Tallgren, the disapproval communicated by international criminal justice ‘risks being unclear or having adverse connotations, depending on the background of the offender’. The same is true for victims and local communities.

Aside from concerns about expression, it is also regrettable that there are no substantive restraints on the scope of international modes of liability – the open-ended system of differentiated ‘modes of liability’ does little to ensure that the standards that courts apply accord with any conceptual foundations. This, as we have seen, is an acute problem when international courts draw so heavily on national doctrine that may or may not accord with basic principles in blame attribution, and when universal jurisdiction allows all range of courts to hear these cases no matter their cultural distance from standards familiar to participants in the trial. In the face of this reality, a unitary theory of perpetration might better preserve (and advertise) culpability as the benchmark for international criminal responsibility, ending the various phases of international experimentation with domestic modes of liability and shifting these academic debates to issues of sentencing, where they belong.

Equally importantly, the unitary theory would simplify a body of rules governing international modes of liability that has attained a degree of technicality that is in jeopardy of alienating those who matter most. From experience, very few practitioners of international criminal justice understand the full import of ‘modes of liability’, which they tend to contract out to experts trained in relevant national jurisdictions wherever possible. This tendency is exacerbated when leading texts describing prominent international modes of liability are not available in official UN languages, further reinforcing the professional dependence on experts socialized in only a small number of jurisdictions. If this impinges on the day-to-day running of international trials, it goes without saying that these issues are likely to further inhibit the engaged participation of victims, perpetrators, and affected communities, who are generally even less equipped to deal with technocratic jargon than the professionals who represent them. It would probably be easier to endure these struggles if they were conceptually unavoidable but of course they are not.

A unitary theory might also mediate the dissonance between national and international concepts of blame attribution more meaningfully. At present, standards of blame attribution vary wildly from one jurisdiction to the next, producing a fragmented array of rules. Predictably, the dissimilarity in international versus domestic standards has caused (and will continue to cause) major practical problems. In one case tried within the Netherlands, for instance, a Dutch court invested considerable energy in determining whether it was required to apply international or domestic notions of complicity when prosecuting its own national for genocide.Footnote ²⁴¹ From the reasoning in the decision, the question appeared determinative of the defendant's responsibility for a crime no less than genocide – application of the national standard of complicity led to conviction; the international equivalent did not. The case highlights how without common standards of blame attribution, serious criminal responsibility presently turns on largely arbitrary elections between two competing notions of blame attribution.

The system is also highly inefficient. As is well known, the two ad hoc UN international tribunals alone are estimated to have claimed roughly 15 per cent of the UN annual budget, which is a projected cost of around $25 million per case.Footnote ²⁴² While it is nigh on impossible to quantify the portion of that figure attributable to the unsettled pluralistic nature of international modes of liability, there can be little doubt that radically limiting litigation over these concepts would free up considerable capacity, save donors resources, and hasten trials.Footnote ²⁴³ Just a short glance at the number of appellate cases that involve complex (but conceptually unnecessary) questions about modes of liability confirms as much. A more efficient system promotes rights to expeditious trial that are frequently in jeopardy internationally and makes capital available that might minimize the selectivity of trials. As such, a more streamlined concept of perpetration promotes accountability.

Finally, to return to one of the central themes of this paper, a unified theory of perpetration is important from a purely functional perspective. Throughout, much of the debate about ‘modes of liability’ has assumed a false duality between mastermind and physical perpetrator. In reality, there are also accomplices who make important (sometimes indispensable) contributions to the ways in which atrocities unfold. I have in mind corporations – the suppliers of weapons, the banks who finance military offensives, representatives of extractive industries who bankroll warlords – that all play surprisingly important roles in sustaining modern bloodshed. Inputs of these sorts have received little more than hortatory acknowledgement in modern international criminal justice, but as soon as this veil is lifted we will see that unified standards of blame attribution are essential in creating a level playing field capable of treating accomplices equally.

Without this, the system of international criminal law enables safe havens, corporate races to the regulatory bottom to avoid liability, and perceptions that businesses in certain jurisdictions are at a competitive disadvantage vis-à-vis those elsewhere. In other areas, international law has some great experience in erecting universal standards in order to deal with these global realities – in treaties ranging from the Warsaw Convention for the Unification of Certain Rules relating to International Carriage by Air to the United Nations Convention on Contracts for the International Sale of Goods, states have rationalized a single set of standards to address transnational practices. So, while I am sensitive to the compelling arguments for pluralism in international criminal justice,Footnote ²⁴⁴ I can only assume that in international criminal justice too the need to avoid overt injustice trumps the otherwise understandable desire for doctrinal heterogeneity between legal systems.

How, then, would this be achieved? A unitary theory of perpetration for international crimes cannot simply replace all standards of attribution everywhere – the prospect of revolutionizing global standards of attribution is politically unthinkable and culturally undesirable. Nonetheless, national courts prosecuting international crimes could use an international unitary standard of perpetration in domestic cases involving international crimes, leaving habitual modes of attribution to continue unaffected for everyday domestic crimes. As a matter of ironic coincidence, this would emulate an extant scheme in German law, which applies a unitary theory of perpetration to a specific subset of administrative offences, even though it maintains its famous differentiated model for other crimes.Footnote ²⁴⁵ The only difference would be that this model would displace and annul customary international standards, then apply uniformly throughout all courts capable of exercising jurisdiction over international crimes.