2. From JCE to indirect perpetration (background)

Central to the notion of criminal liability in Nuremberg was the idea that membership in certain organizations was itself criminal.Footnote ⁵ The International Criminal Tribunals for Yugoslavia (ICTY) and Rwanda (ICTR) rejected the concept of membership in criminal organizations as a mode of liability. The discussions leading up to the ICTY statute emphasized individual rather than collective responsibility. The UN Secretary-General commented on this shift in a letter to the President of the Security Council:

The fact remains, however, that the majority of crimes in international conflicts are carried out by (mostly large) groups and as a matter of organizational policy and not by individuals on their own initiative.

In addressing this problem of collective responsibility, the ad hoc tribunals employ the concept of JCE. There are three types of JCE that have emerged in the tribunals’ jurisprudence. In order to be liable for the basic form of JCE (Type 1), participants must engage in the criminal acts committed by the group with the requisite mens rea, which means the individual members must share the intent of engaging in the particular conduct and/or bringing about the particular results.Footnote ⁷ The second form of JCE (Type 2) requires that participants take part in a system of mistreatment (like a concentration camp) while aware of the nature of the system and with intent to further it.Footnote ⁸ The final type of JCE (Type 3) is the most controversial because it imposes liability for collateral offences if these are foreseeable and the participant is reckless as to the possibility of bringing them about.Footnote ⁹ Although I cannot consider the controversy regarding Type-3 in detail here, it is important to note that it has been used by the ICTY to impose liability on the leaders of a criminal enterprise.Footnote ¹⁰ For the purposes of my analysis here, it is sufficient to compare indirect perpetration with the less controversial forms of JCE (Types 1 and 2).

From its first trial in 2009, the ICC declined to use JCE for attribution of responsibility to high-level perpetrators who use a criminal organization to carry out the crimes. Instead, it relied on the mode of liability articulated in Article 25(3)(a) of the Rome Statute. Article 25(3) reads in relevant part:

In the Lubanga case (the ICC's first), the ICC judges interpreted this provision to include co-perpetration by means of control over crime, while in Katanga and Ngudjolo, the judges further explained that control over crime can consist in organizational control.Footnote ¹¹ With this move, the ICC adopted co-perpetration by means as a way to charge leaders for crimes committed by their subordinates. The most recent application of the concept by the ICC prosecutor in the application for the arrest warrant of the President of Sudan, Omar Hassan Ahmad Al Bashir, gave rise to extensive scholarly discussion.Footnote ¹² My goal in taking up this controversy is to address the theoretical strength of the indirect (co-)perpetration model as a way of establishing the responsibility of higher-up commanders for the acts committed by their subordinates. As the first step in this analysis, I turn to the origins of the concept that has become central to the ICC's concept of criminal liability.

3. Roxin's theory of indirect perpetration

The best starting point for the discussion of perpetration by means is Roxin's own articulation of the idea in the early 1960s. A segment of Roxin's seminal essay was recently published in English for the first time in the Journal of International Criminal Justice along with a collection of articles devoted to discussing its meaning and implications.Footnote ¹³ These passages contain the foundations of the theory of liability now being applied by the ICC, so I begin by presenting its main points. Roxin defined perpetration as ‘domination of the act’ (Tatherrschaft). ‘A perpetrator, Roxin held, is a person who ‘dominates’ the commission of the criminal offence in that he or she has the power to determine whether or not the relevant acts are carried out’.Footnote ¹⁴ Roxin's overall aim was to distinguish principals from accessories, reflecting the common intuition that the former are more culpable in bringing about criminal harm than the latter.

Roxin was concerned that German courts presiding over prosecutions of Nazi crimes in Germany in later 1940s and 1950s were relying on a subjective theory to make the distinction. If subjective intentions alone set perpetrators apart from those who merely assist in the commission of the crime, then it becomes much easier to argue that an accused does not share the intention of the perpetrators but merely wishes to assist them (perhaps because she did not want to disobey superiors, and/or was afraid to be harmed or lose a job as a result of disobedience). Assistance entails a lesser degree of culpability precisely because the requisite intention appears to be missing. As a result, a conviction for aiding and abetting rather than perpetration may (although it does not necessarily) prompt judges to consider mitigating factors.Footnote ¹⁵ In postwar Germany, the subjective theory served the unseemly purpose of minimizing individual responsibility of anyone but Hitler and the top Nazi command and bolstered revisionary assessments of the Nazi era in which many ordinary Germans were all too happy to view themselves as Nazism's victims rather than its enablers.Footnote ¹⁶

In his contribution to the symposium on indirect perpetration, George Fletcher discusses the corresponding distinction between principal and accessory in American law. He writes:

I will return to Fletcher's point about the need for ‘principled distinctions’ in Section 6 below, but it is important to note here that Roxin's defence of indirect perpetration is, to a large extent, motivated by the need to provide accurate labels for the modes of liability employed by the criminal justice system. These labels are accurate only if they reflect the degree of personal culpability properly attributable to the particular defendant.

According to Roxin, the issue of indirect perpetration is particularly pressing for international criminal law because international crimes, unlike domestic crimes, often tend to be perpetrated by groups rather than individuals ‘or a few people on the basis of independent initiative.’Footnote ¹⁸ Difficulties arise when criminal law concepts created to deal with individual acts need to be applied to crimes committed by groups and organizations, often ones that involve the state apparatus itself. Jens Ohlin provides a stark formulation of this challenge in his article ‘Joint Intentions to Commit International Crimes’ that I will discuss in detail in Section 5. In his words: ‘The question is how to develop a sophisticated doctrine that navigates between the collective nature of international criminality and the individualized determinations of criminal law’.Footnote ¹⁹ Ohlin criticizes the ICC's decision to embrace Roxin's solution to the problem as shifting the balance too far in the direction of act, away from intention. However, Ohlin's own theory of joint intentions suffers from the opposite fault of overemphasizing collectivity (see Section 4 below). First, however, what exactly is Roxin's solution?

Two defendants serve as Roxin's starting point. One is Adolf Eichmann whose lawyer argued at the trial in Jerusalem that his client's ‘crime is not the act of an individual; the state itself is the perpetrator’.Footnote ²⁰ The other is Bogdan Stashynsky, a KGB agent, who assassinated two Ukrainian nationalists in Germany (Lev Rebet in 1957 and Stepan Bandera in 1959).Footnote ²¹ The Federal Supreme Court in Germany found Stashynsky to be an abettor of the murders because he had no interest in committing the crimes but was a mere tool of the KGB. Roxin argues that neither Stashynsky nor Eichmann can meaningfully be called an abettor. The former committed the act which the KGB orchestrated. Stashynsky, the agent, and KGB leadership that ordered the assassination are perpetrators – direct and indirect. Eichmann, in turn, committed crimes through an organization in which he held a senior position. The key to Roxin's position is that in these types of collective crimes both the behind-the-scenes perpetrator and the actual direct perpetrator are principals rather than mere accessories. In contrast, domestic law conceives of perpetration by means in terms of an innocent person being used as an instrument to commit a crime.Footnote ²² The tension here is apparent: How can there be a fully liable perpetrator behind the guilty actor? How can responsibility for the same act be attributable to two distinct people who did not commit the crime together?

Roxin answers by relying on the idea of Tatherrschaft (control over the act).Footnote ²³ An indirect perpetrator controls the acts of a culpable actor by means of an organized power structure.Footnote ²⁴ Within this structure, a particular (fully culpable) perpetrator does not alone control the occurrence of the crime. Instead, those in charge of the organization can replace actors with others and still make sure that the crime is committed. A complex criminal organization like the Nazi extermination machine continues to operate regardless of whether or not a particular actor carries out his part. This exchangeability does not, of course, relieve individuals of criminal responsibility (contrary to what Eichmann tried to argue). Instead, it turns the actions of superiors who set the machine in motion, but do not necessarily commit the particular murders, into perpetration.Footnote ²⁵ Roxin writes:

It is important to note that perpetration by means can be attributed to behind-the-scenes actors in various positions of authority. In other words, not only leaders but anyone whose position within the organizational structure gives him or her control of the crime is liable as a principal. Eichmann, for instance, was not at the very top of the organizational hierarchy, but he did exercise control over the acts of his subordinates. In this way, at each level of the hierarchy, a perpetrator can be both a fully culpable instrument and an indirect perpetrator.

4. Indirect perpetration and the German courts

The German Supreme Court applied Roxin's concept of perpetration by means in the 1994 trial for the killings of East Germans who tried to cross the border into West Germany. The Court ascribed criminal responsibility to the high-ranking accused. Keßler, Streletz, and Albrecht were members of the National Defence Council for the German Democratic Republic, the organization in charge of establishing the border policy of killing those attempting to cross. The District Court found Keßler and Streletz liable as instigators (Anstifter) and Albrecht as an aider and abettor (Gehilfe) for the killings carried out by the border guards.Footnote ²⁷ However, the Federal Supreme Court of Germany rejected these verdicts, applying the category of indirect perpetrator instead.Footnote ²⁸ It may be significant for evaluating the status of the concept in international customary law that the Court did not simply rely on Roxin's theory but rather cited a number of different scholars who concurred that ‘indirect perpetration was possible despite the direct perpetrator's criminal liability’.Footnote ²⁹

The Court focused on the element of control over crime, which belongs fully to the perpetrator and not (most of the time) to the accessory. Cases in which the decisions of the behind-the-scenes actor lead to the commission of the crime serve as examples of indirect perpetrations. The high-level leaders who order the crime are not accessories of the direct perpetrators; rather, they control whether or not the act is going to take place. The border guard cases belong to this category. The following passage is a concise summary of the circumstances creating the conditions for indirect perpetration:

The control theory provides a good explanation for the way behind-the-scenes perpetrators can operate through those who physically commit the crime. As did the Jerusalem Court in the Eichmann trial, the German Court here emphasized that distance from the deed can in these situations increase rather than decrease the degree of responsibility.Footnote ³¹ Roxin's theory, together with its application in the trials of high-level official for the German border killings, set the stage for the ICC's use of the concept of perpetration by means. The one significant ‘blind spot in the argument’Footnote ³² of the German Supreme Court in the border case is the failure to consider that the defendants, along with others in the GDR defence council leadership perpetrated the crimes together. The ICC corrects this oversight by applying the concept of indirect co-perpetration.

5. A new alternative? Ohlin's theory of joint intentions

In his critique of the perpetration by means doctrine Ohlin labels it ‘the new darling of the professoriate’.Footnote ³³ Contrary to the endorsements that the ICC theory has won, Ohlin argues that it represents an overreaction to the shortcomings of the JCE doctrine. What is missing from the ICTY analysis of JCE, Ohlin claims, is the notion of joint intentions. The doctrine of indirect (co-)perpetration does not address this shortcoming. Instead, it focuses on the control that the indirect perpetrator exercises over the acts carried out by others. Admittedly, the indirect (co-)perpetration doctrine does address a central problem with JCE, namely the inadequate link it establishes between the actions of the leader (the behind-the-scenes perpetrator) and those of the subordinates. This weakness leads to the expansion of JCE to include a recklessness standard – vicarious liability for foreseeable but not intended results. Many commentators find this extension troubling, but Ohlin does not think that we need to abandon all the elements of JCE to remedy the problem. Instead, he advocates the joint intentions model of collective responsibility; this model is grounded in a philosophical theory of joint intentions.

Ohlin's shared intentions theory draws on the philosophies of Bratman and Searle, among others. The key relevant feature of these accounts is that the collective nature of acts committed by a group of people lies in their ability to plan and co-operate with regard to the results. Co-operative behaviour is usually characterized, Ohlin argues, by pooling information and shared decision-making.Footnote ³⁴ However, it is also possible in the absence of deliberation.Footnote ³⁵ Ohlin then goes on to apply the shared intentions theory to five hypotheticals in order to demonstrate that it provides better outcomes than the control theory. His hypotheticals include the love parade stampede, the Essen lynching, the Concentration camp, the deportation, and the attack against civilians. Very briefly, the first is based on the stampede at a concert in Germany in which many people were killed in a crowd after they got trapped in a narrow tunnel leading to the outdoor location of a concert. The second is the lynching of British prisoners of war by a crowd in Essen at the instigation of a German officer. In the third example, three officers work as guards at a concentration camp in the former Yugoslavia. They are aware of the murder, torture, and rape systematically committed at the camp. One of them commits torture and rape himself, while the others do not and are not even aware of the particular incident. Fourth, a unit of soldiers is engaged in deportation as part of an ethnic cleansing campaign. One of them decides to shoot the victims instead of proceeding with the deportation. Finally, a unit of soldiers ordered to seize a town decides to kill all the civilians.Footnote ³⁶

Ohlin is surely correct in saying that the joint-intentions theory accurately captures the various levels of liability involved in these five cases that range from deaths resulting from a crowd that rushes to escape a fire to intentional killings. Furthermore, this analysis has important consequence for the JCE doctrine. It imposes two restrictions on JCE: the first one consists in ‘eliminating vicarious liability for actions that fall outside the scope of the original agreement’, the second in ‘differentiating levels of participation in the group endeavor’.Footnote ³⁷ But to say that the control theory is less successful at explaining liability in the five cases above is not to say that it does not provide a valuable form of criminal liability. Indeed, the joint intentions theory raises some of the same problems that Roxin saw in the subjective approach to the perpetrator/accessory distinction. Ohlin explains:

This explanation brings us back to the problems of applying a subjective analysis to cases like Eichmann or Stashynsky. If defendants in those cases claim not to share the intention of the behind-the-scenes perpetrators (Heydrich or Hitler in Eichmann's case, the KGB leadership in Stashynsky's case), can they be considered accessories? For Stashynsky, the answer is simpler – the actual perpetrator is not an accessory. For Eichmann, organizational control over crime provides the answer. None of the five cases discussed by Ohlin involve this organizational structure central to Roxin's (and ICC’s) theory of indirect perpetration. As a result, his criticism appears to be misplaced. The control theory provides an objective way to distinguish perpetrators and accessories that the intentional theory lacks. The focus on organizational structure may pose problems for the application of indirect co-perpetration to cases where bureaucratic organization is less apparent.Footnote ³⁹ Despite the potential evidentiary challenges, however, this form of liability provides a particularly fitting way to charge those responsible for the most serious international crimes.

In sum, while a nuanced analysis of joint intentions is indeed valuable, I do not think that it can solve the problems that JCE presents. Rather, the control theory is much better at dealing with these issues. More specifically, it better reflects culpability by setting aside a separate category for the indirect perpetrators who ensure the commission of crimes from a distance.

Neither JCE nor Ohlin's joint intentions theory captures the higher degree of responsibility belonging to leaders of criminal organizations as effectively as the concept of indirect perpetration does.Footnote ⁴⁰ In fact, Ohlin identifies this failure of JCE, making the following observation:

Ohlin's solution is to replace JCE by what he calls ‘two more specific and accurate modes of liability: (1) co-perpetrating a JCE; and (2) aiding and abetting a JCE.’Footnote ⁴² Co-perpetration would be a mode of liability appropriate for actors satisfying three conditions: ‘(1) they participate in a joint criminal endeavor at a high level; (2) they have the intent of furthering the criminal purpose of the group endeavor; and (3) they are indispensable to the success of the joint criminal endeavor’.Footnote ⁴³ While this approach focuses specifically on the intent, it fails to provide sufficient differentiation for high-level perpetrators. Ohlin acknowledges that the theory requires further elaboration of the meaning of ‘high level’ participation:

But this is a crucial point: to focus the distinction between perpetrator and abettor primarily on the intention behind the act does not do justice to the differences in degree of participation between the mastermind of the act and the mere foot soldier.

Before turning to a more explicit defence of indirect perpetration, one last point regarding the need for differentiation of liability must be addressed. While few disagree that there is indeed a difference between a high-level participant who orders the crimes and the foot soldier who carries out the orders, some argue that the difference can be addressed at the sentencing stage rather than by applying distinct forms of liability.Footnote ⁴⁵ However, I think Ohlin is right to claim that ‘it seems intuitively correct that the culpability principle is about more than just punishment’.Footnote ⁴⁶ In this regard, a recent article by Werle and Burghardt entitled ‘Establishing Degrees of Responsibility: Modes of Participation in Article 25 of the ICC Statute’ provides a useful starting point. The authors argue that Article 25(3) of the ICC Statute favours a differentiation model of liability for the following three reasons: (i) the literal and contextual reading of the statute itself; (ii) the case law of the ad hoc tribunals; and (iii) the ‘phenomenological and normative features of international criminal law and the aim and purpose of [Article] 25(3) of the ICC Statute all support the differentiation model’.Footnote ⁴⁷ Although a detailed discussion of these arguments belongs elsewhere, it is worth noting that one reason for taking the control theory seriously is that it captures the significance of liability as such, apart from any practical consequences it may have for sentencing.

Even leaving the debate regarding modes of liability aside, it should be fairly uncontroversial to say that a way must exist to charge leaders who lack actus reus as principals. The argument that I advance here is that indirect perpetration does the job better than the available alternatives (including Ohlin's revision of JCE) because it achieves a more balanced mix of objective and subjective criteria.

6. Preserving the balance between intention and act

To define the principal in a criminal act as the agent who has control over crime is to concede the moral intuition that people should be responsible for things that they choose to do. In this way, it is not a purely objective test. Rather, control is intimately bound up with the intention to see the act carried out. In order to mount a defence of indirect (co-)perpetration against the criticisms proposed by Ohlin, it is necessary to articulate the elements of this doctrine in some more detail. Although these elements are by no means firmly settled yet, the Rome Statute and the ICC have formulated some of the key criteria. In the remainder of the article, I present these elements and elaborate on the primary advantages of the concept of indirect perpetration. These advantages are consistent with Roxin's original reasons for proposing the theory: the concept is so valuable for international criminal law because it allows for an objectively grounded distinction between perpetrators and accessories, ascribing liability to those who order crimes but do not themselves participate directly in their commission.

Article 30 (‘Mental Element’) of the Rome Statute provides for the subjective criteria that must accompany any of the acts described in Article 25 as preconditions for a finding of liability:

1. 1. Unless otherwise provided, a person shall be criminally responsible and liable for punishment for a crime within the jurisdiction of the Court only if the material elements are committed with intent and knowledge.

2. 2. For the purposes of this article, a person has intent where:

   1. (a) In relation to conduct, that person means to engage in the conduct;

   2. (b) In relation to a consequence, that person means to cause that consequence or is aware that it will occur in the ordinary course of events.

3. 3. For the purposes of this article, ‘knowledge’ means awareness that a circumstance exists or a consequence will occur in the ordinary course of events. ‘Know’ and ‘knowingly’ shall be construed accordingly.

In accordance with these requirements, anyone convicted as an indirect perpetrator would need to have intent and knowledge. Intent in relation to conduct is entirely volitional – that person means to engage in the conduct – while intent with regard to consequences may be either volitional or cognitive – the person is ‘aware that it will occur in the ordinary course of events’. As Heller points out in his analysis of the Rome Statute, it is unclear whether the latter includes recklessness as well.Footnote ⁴⁸ Furthermore, the ‘otherwise provided’ clause refers to some crimes that require specific intent under the statute. Paragraph (a) of Article 25(3) establishes criminal responsibility for principals through either direct or indirect commission, while paragraph (b) covers inducement, paragraph (c) aiding and abetting, and paragraph (d) other types of accessory liability.Footnote ⁴⁹ By these divisions, Article 25(3)(a)–(c) codifies the distinction between principals and accessories, which Roxin aimed to establish through his doctrine of indirect perpetration.Footnote ⁵⁰ Although the intention requirement remains crucial under Article 30, the distinction is grounded in the objective control requirement for indirect perpetrators.

The ICC defined the objective elements in the Katanga case. In brief, they are as follows:

1. 1. ‘The organization must be based on hierarchical relations between superiors and subordinates’;

2. 2. ‘The organization must have enough subordinates so that the carrying out of orders becomes “automatic”’ (the Court quotes Roxin on this point, as well as many others);

3. 3. ‘The indirect perpetrator must exercise control and this control must be manifest in the subordinates’ compliance (evidence of control includes capacity to hire, train, impose discipline, and provide resources to his subordinates)’; and

4. 4. ‘Compliance includes commission of crimes that are under ICC jurisdiction’.Footnote ⁵¹

The evidentiary burden for proving indirect perpetration is quite high – appropriately so, given the higher degree of liability belonging to principals. However, such an increased burden should not lead the prosecutor to avoid indirect perpetration and opt for a different mode of liability altogether (such as command responsibility under Article 28, for example). While perhaps easier to prove, superior liability, especially for civilians, focuses on omissions rather than commissions. Furthermore, Article 25 should not be bypassed because it is the primary source of individual responsibility under the Statute.Footnote ⁵²

The ICC and German courts are not alone in using the concept of indirect perpetration to convict leaders who commit crimes through an organized crime apparatus. This wider use may lend some additional legitimacy to the concept. The symposium on indirect perpetration in the Journal of International Criminal Justice includes two articles on domestic use of the concept. Kai Ambos’ contribution deals with the 2009 conviction by the Peruvian Supreme Court of the former president of Peru, Alberto Fujimori. The Court convicted Fujimori for indirect perpetration of crimes against humanity during his presidency by means of an organized power apparatus. The Court followed Roxin's criteria including the presence of a hierarchical organization, the actual exercise of command by the indirect perpetrator, and the interchangeability of the direct perpetrator.Footnote ⁵³ Ambos concludes that its use in the Fujimori conviction demonstrates that the theory of indirect perpetration can be used successfully to prosecute serious international crimes. Another article in the symposium by Francisco Muñoz-Conde and Héctor Olásolo, suggests that indirect perpetration has come to play an important role in a number of Latin American jurisdictions.Footnote ⁵⁴ The authors examine instances in which courts of various Latin American countries have used (or rejected) the concept of indirect perpetration and conclude that it has gained some ground in recent years, even if it remains controversial. This comparative overview is valuable in showing that courts find the concept of indirect perpetration useful in the context of prosecuting leaders of organized power structures for serious international crimes like large-scale human rights violations and crimes against humanity.

Indirect perpetration provides international criminal law with an instrument for capturing a proper level of liability. Culpability generally reflects the blameworthiness of what an actor has done and can include, as the idea of moral luck described in the introduction illustrates, more than just intentions.Footnote ⁵⁵ This sense of culpability is particularly difficult to capture adequately in international crimes because of their collective nature. As many have noted, international crimes

In this moral ‘gray zone’,Footnote ⁵⁷ a legal doctrine reflecting culpability is both especially important and especially difficult to articulate.

The idea that crimes must be labeled fairly to reflect harm as well as culpability is firmly established in criminal law theory. ‘Although not an element of any crime per se, fair labeling nevertheless is viewed as a fundamental principle of criminal law: that “justice not only must be done, but must be seen to be done”’.Footnote ⁵⁸ The source of this principle is a subject of some debate, which I can only mention in passing here. While some believe that it is the nature of criminal law doctrine itself which requires that various degrees of culpability be proportionately represented by corresponding labels,Footnote ⁵⁹ others have argued that there exists a basic human right to fair labeling.Footnote ⁶⁰ Whatever the ultimate justification for fair labeling, the requirement of fairness cannot be explained with reference to case law or statutory provisions alone. As Ohlin suggests, for example, ‘moral and philosophical theories that ground culpability are crucial, particularly when courts are . . . in the position of not just applying, but also of announcing such theories’.Footnote ⁶¹

In his analysis of indirect perpetration, George Fletcher also advocates a proto-philosophical grounding for categories of liability. More specifically, he sees the turn to Roxin's theory by the ICC as indicative of a return to an old German Dogmatik (roughly defined as a supporting culture of ideas and principles). He applauds this turn because it is motivated by a concern with individual rather than restorative justice. Fletcher concludes his essay with the following passage:

Of course, the focus on individual justice should not be overstated. As I have discussed, collective responsibility problems dominate the theories of responsibility in international criminal law. However, the renewed attention to the perpetrator's acts, rather than the consequences of those acts, provides an important reminder that individual liability must be grounded with precision, precision that many have argued is lacking in the JCE doctrine.

Indirect perpetration is better at fair labeling in the context of serious international crimes because it achieves a more adequate balance between the focus on intention and that on act. While not applicable in every context, this theory of liability is particularly appropriate for establishing the responsibility of commanders (especially those at a high level) for crimes they commit by means of an organized group of direct perpetrators. By focusing on the control that the superiors exercise of the direct actors, indirect perpetration (and co-perpetration) reflects the moral intuition that a commander's criminal actions are highly culpable and gives expression to this intuition in criminal jurisprudence.