2.1. Control-theory and joint perpetration

With respect to joint perpetration, the Pre-Trial Chamber in Lubanga defined ‘control’ as ‘joint control over the crime by reason of the essential nature of the various contributions to the commission of the crime’.Footnote ¹¹ The Chamber recognized that in cases of joint perpetration none of the perpetrators normally ‘controls’ the commission of the offence by himself, because the defining feature of co-perpetration is a division of labour.Footnote ¹² But

The defining feature of joint perpetration, according to the Chamber's definition, is a hypothetical power, namely ‘the power to frustrate the commission of the crime by not performing their tasks’.Footnote ¹⁴ Because this decisive criterion is framed in negative terms (‘frustrating’ by ‘not performing’), there exists no particular affirmative act that a person must perform in order to become a joint perpetrator. The Lubanga Trial Chamber indeed emphasized that a person can be a co-perpetrator even where he does not physically perpetrate any of the elements of the crime in question and where he is not even present at the scene of the crime.Footnote ¹⁵ It can be sufficient, for example, for the actor to be a ‘mastermind’ who decides ‘whether and how the offence will be committed’.Footnote ¹⁶

The effect of the control-theory, as devised by the Lubanga Pre-Trial and Trial Chambers, is ambivalent. The theory limits the concept of (joint) perpetration to those participants in a criminal enterprise whose contribution is a condition without which the criminal plan could not have come to fruition. But on the other hand, the theory expands the scope of perpetratorship to persons who are far removed from the scene of the crime and do not personally perform any of the acts required by the offence definition.

2.2. Criticism of Judge Fulford

Judge Fulford, in his separate opinion in Lubanga, agrees with the expansive element of the definition given by the majority of the Trial Chamber. But he regards that definition as too narrow where it demands an ‘essential’, indispensable contribution. Judge Fulford proposes instead a very simple test for joint perpetration, namely whether there exists ‘an operative link between the individual's contribution and the commission of the crime’.Footnote ¹⁷ By applying this test, a court would be able to avoid ‘a hypothetical investigation as to how events might have unfolded without the accused's involvement’.Footnote ¹⁸ Yet, Judge Fulford seems to equate the required ‘operative link’ with causation, because in the following paragraph of his opinion he demands for co-perpetration that there exist ‘a causal link between the individual's contribution and the crime’.Footnote ¹⁹ That phrase raises the question what exactly Judge Fulford understands by a ‘causal link’. If an act (say, the furnishing of a weapon to a murderer) is ‘operative’ in the commission of the offence (because the murderer uses that weapon), does that create a causal link between the furnishing of the weapon and the killing? If so, how would Judge Fulford distinguish between a (joint) perpetrator and a mere aider and abettor? Can there be different (stronger or weaker) types of ‘causal link’? Can there be a ‘causal link’ that is not essential? Does causation not necessarily imply that without the existence of the factor in question the consequence would not have occurred? And if not, why would a marginal, easily replaceable contribution be sufficient to turn a mere helper into a co-perpetrator? How about a man who provides not the weapon used in the offence but a bicycle which the killer rides to the site of the crime? Would furnishing the bicycle also have an ‘operative’ or ‘causal’ link to the killing? Would the answer to that question depend on whether the bicycle was, under the circumstances, the only means by which the killer could arrive at the relevant site in time to kill the victim? If so, does Judge Fulford's analysis not also have to take into account whether the means provided was ‘essential’?Footnote ²⁰

There may well be convincing answers to all these questions, but Judge Fulford unfortunately does not provide them. The test of perpetratorship which he suggests therefore remains vague and leaves the judges very much to their intuition rather than providing them with standards by which to make the difficult distinction between perpetration and mere accessorial liability. Judge Fulford is correct, of course, in pointing out that many legal systems, including the Statute of the ICC, do not provide for different sentencing levels for perpetrators and accessories. Yet, as long as the law distinguishes between several forms of involvement in a criminal offence – and the ICC Statute in Article 25 (3) clearly makes such a distinctionFootnote ²¹ – those who apply the law may not use the various labels arbitrarily. A judge cannot on Monday convict a defendant of aiding and abetting, and on Tuesday convict another defendant of joint perpetration on the same or very similar facts, telling the Tuesday defendant that it makes no difference, in the result, whether he is convicted of perpetration or of aiding and abetting.Footnote ²²

2.3. Criticism of Judge Van den Wyngaert

Judge Van den Wyngaert, in her concurring opinion in Ngudjolo, takes issue with both effects of the control-theory; in her opinion, that theory is at the same time too broad and too narrow. She suggests that joint perpetration does not require an ‘essential’ causal contributionFootnote ²³ but ‘a direct contribution to the realisation of the material elements of the crime’.Footnote ²⁴ Like Judge Fulford, Judge Van den Wyngaert rejects the ‘essentiality’ requirement because it compels judges ‘to engage in artificial, speculative exercises about whether a crime would still have been committed if one of the accused had not made exactly the same contribution’.Footnote ²⁵ But she also finds insufficient Judge Fulford's broad-stroke approach under which anyone who provides some ‘causal’ element can be treated as a perpetrator. Instead, Judge Van den Wyngaert would limit perpetratorship to those who directly bring about the material elements of an offence.Footnote ²⁶ She concedes that the notion of ‘direct’ perpetration is not easy to apply to some of the more complex offences typical of international law, such as displacing a civilian population in violation of Article 8(2)(e)(viii) of the ICC Statute. In these cases – which may well be the majority of cases coming before the ICC – she would regard as ‘direct’ perpetrators even those who plan or organize the acts in question, because planning is ‘an intrinsic part of the actual execution of the crime’.Footnote ²⁷

One may regard this ‘softening’ of Judge Van den Wyngaert's approach as a sign of welcome flexibility. But the adaptations proposed by Judge Van den Wyngaert also indicate that the criterion of ‘directness’ has only a modest measure of distinctive substance. If ‘direct’ causation can also mean participation in the planning stage long before the actual displacement of civilians takes place (to use Judge Van den Wyngaert's example), what then is the difference between ‘direct’ and ‘indirect’ or ‘remote’ causation? Judge Van den Wyngaert claims that what is ‘direct’ will have to be determined by the facts of each case.Footnote ²⁸ But if it is the judges who in each individual case need to determine what is ‘direct’ and what is not, why not simply leave the question who is a perpetrator to the appreciation of the court, as Judge Fulford suggests?

The ‘directness’ criterion also lacks a convincing normative basis. Judge Van den Wyngaert's main argument in favour of this criterion is that it treats individual and joint perpetrators equally: since a person acting alone can be convicted only if he ‘brings about’ the material elements of the offence (e.g., by shooting the victim), she claims, the same should apply if two or more persons act jointly.Footnote ²⁹ But joint perpetration differs in one critical aspect from perpetration by an individual offender: the job of committing the offence is divided up among the co-actors; they join forces for the very purpose of relieving each participant of the necessity to ‘bring about’ by himself the result of the criminal plan. If it were a necessary requirement for every joint perpetrator to individually fulfil each element of the offence definition, the concept of joint perpetration would be superfluous – every participant could be convicted as an individual perpetrator. Judge Van den Wyngaert's equation of individual and joint perpetration thus misses the very point of joint perpetration: the division of labour among the co-perpetrators.

2.4. What makes joint perpetration?

What, then, is the distinctive feature of joint perpetration? Before we proceed to suggest an answer to this question let us consider why it is necessary to do so. As we said above, whenever the law differentiates between perpetrators and accessories there have to be criteria for drawing the line between these categories, lest the courts apply them on a mere hunch or on arbitrary grounds. A need to define joint perpetration and to keep it apart from other types of criminal liability therefore exists in all legal systems that, like the ICC Statute, distinguish between perpetrators and accessories.

As we have seen, single-word ‘theories’ (relying on expressions like ‘direct’ or ‘operative link’ to solve the problem) hardly satisfy the need to inform and shape the application of the law in this area, because they do not offer more than a very general idea about what may be the gist of joint perpetration. The control theory as devised by the Lubanga Pre-Trial Chamber has the great advantage of offering a binary criterion: either one's contribution to the completion of the crime is essential, i.e., a conditio sine qua non – or it is not. And the ‘essentiality’ criterion is also prima vista plausible: it reflects the value judgement that those who provide central contributions are, in general, more blameworthy than those who remain at the margins of the criminal enterprise and provide only support that the main actors could have done without.

Yet, even the control-theory may lack sufficient sophistication. One of its problems is the question from what perspective the ‘essential’ character of a given act is to be determined. This can be done from an ex ante and an ex post perspective; that is, looking forward from the stage of planning the crime,Footnote ³⁰ or looking backward after its completion (or frustration). The analysis of the ‘essentiality’ of any participant's contribution is easier and more reliable if one gauges it ex ante, for example at the planning stage of the crime. Looking at the plans of the persons involved, it is not difficult to determine whose contribution they deem indispensable and which contributions they regard as ‘accessorial’; that is, useful but not crucial to the success of the plan. Taking an ex ante view has the disadvantage, however, that the judge must accept as binding any miscalculation on the part of the conspirators: if A and B think that a robbery can succeed only if B provides a gun, that expectation would, under an ex ante ‘essentiality’ test, turn B into a joint perpetrator if he actually furnishes the gun; and that would hold true even if A does not use the gun in the robbery but obtains the money by mere verbal threats. Using an ex post perspective (‘Was B's contribution “essential” to the robbery as it was actually carried out?’) avoids this difficulty but leads to the problem of hypothetical guessing, which Judges Fulford and Van den Wyngaert raised in their opinions. If A, in the robbery hypothetical, actually uses B's gun in order to threaten the victim, we cannot say with any degree of certainty what the outcome would have been if B had not provided the gun. Perhaps A would have desisted from the robbery altogether, or perhaps he might have hidden a banana or some other object in his pocket and pretended that it was a gun, thus threatening the victim sufficiently to make him hand over the money. In conclusion, then, while both perspectives may lead to different results in any given case, neither is without flaws.

But that is not even the main problem with the ‘essentiality’ test. Contrary to what Judge Fulford has written,Footnote ³¹ the Lubanga court's control-theory may not define perpetratorship too narrowly but too broadly. The way the Trial Chamber applies the test to the facts of Lubanga gives an indication of how far the net of perpetration is cast under the seemingly narrow ‘essentiality’ test. A joint perpetrator, the Lubanga majority holds, does not have to be present at the scene of the crime, and there need not even exist a direct or physical link between his contribution and the commission of the crime; it is sufficient for a perpetrator to assist in formulating the relevant strategy or plan, to become involved in directing or controlling other participants' or to determine the roles of those involved in the offence.Footnote ³² If all these contributions, which can be quite remote in time and place from the commission of the offence, are deemed ‘essential’ and thus sufficient to establish perpetratorship, then one must ask what remains for mere accessorial liability as an instigator or an aider and abettor. If, for example, a scientist in 2012 provides critical information which enables the leadership of a state to produce chemical weapons prohibited under Article 8(2)(b)(xviii) ICC Statute, is he then – assuming he has foreseen the use of the weapons – a joint perpetrator of an attack carried out with these weapons in 2014?

If we wish to distinguish between those who are at the centre of the criminal offence and therefore deserve to be labelled as perpetrators, and those who remain at the margins in a supporting role and should therefore be punished as accessories, then the one-dimensional criterion of the indispensability of the contribution may be insufficient because it captures only one aspect that may be relevant. We might indeed have to give up the (attractive) idea of basing the distinction on a single factor and look instead for a cluster of factors that will have to be taken into consideration. In doing so, we should be aware that the line between (joint) perpetrators and accessories cannot be drawn on empirical grounds but requires a normative (value) judgement – a judgement that in the last resort is based on a notion of fair attribution and is therefore soft at the edges. The factors relevant for this value judgment cannot simply be deduced from the facts of each individual case, because facts by themselves give us no clue as to their normative valuation. What we can do, however, is try to identify factors which indicate a ‘central’ role in a criminal enterprise, and thus perpetratorship. Even when we have agreed on such factors, there is still plenty of room for judicial weighing and balancing in each case, because these factors may, on a given set of facts, pull in opposite directions. But at least we would have criteria by which to rationally evaluate any borderline case.

One approach in defining such criteria centres on mens rea. It has often been emphasized that participation in or at least adherence to a common plan is one factor that must exist in any case of joint perpetration. A person who does not cooperate with others on the basis of some – albeit silent – agreement may be liable as an individual perpetrator but cannot be a ‘joint’ perpetrator. What other mens rea requirements may be necessary for perpetratorship depends on each offence definition.Footnote ³³ One should note, however, that a common plan, as that term is traditionally understood by international courts, often exists also between perpetrators and accessories: an instigator consciously works together with the perpetrator in the initial stage, and aiders and abettors also often support the commission of the offence based on an agreement with the main perpetrators. The existence of such a minimal agreement therefore is a necessary but not a sufficient condition of joint perpetration. One of the authors of this article has recently developed the idea that a joint intention that the group commit a collective crime is the defining element of co-perpetration; what is required, under this approach, is joint deliberation and the meshing of sub-plans among co-perpetrators.Footnote ³⁴ Another subjective factor of possible relevance is a strong personal interest in the success of the criminal enterprise, going beyond the minimal requirement of mens rea. Such a personal interest may be regarded as a (weak) indicator of joint perpetration, because a person interested in the outcome will, ceteris paribus, contribute more eagerly and persistently than a person who acts out of altruism or for a set fee.Footnote ³⁵

An alternative approach places greater emphasis on objective factors. If one follows that approach, the indispensable nature of a person's contribution weighs heavily as an indicator of perpetratorship, and might even be regarded as a necessary requirement for a finding of perpetratorship. A person whose contribution has no critical bearing on the implementation of the criminal plan hardly qualifies as a ‘central’ or ‘essential’ participant. But as we have tried to show above, the indispensability of a contribution is not under all circumstances a sufficient condition for attributing perpetratorship. One option is to add an element of immediacy, of carrying out a task temporally close to the commission of the material elements of the offence. Roxin, whose writings have to some extent influenced the international debate on this subject,Footnote ³⁶ has suggested that only those who participate in the crime after the stage of attempt has been reached should be considered co-perpetrators.Footnote ³⁷ This limitation goes in the right direction but may be a bit too ‘technical’ since it would make joint perpetration depend on the vagaries of the definition of attempt. But under an objective approach, an element of ‘control’ over the actual commission of the offence is an important indicator of joint perpetration. Typically, a joint perpetrator (co-)decides whether and how the offence is actually perpetrated, either by directly taking part in the actus reus or at least by overseeing – by telephone, for example – the commission of the crime by the immediate actors.Footnote ³⁸

In sum, the concept of ‘control’ in the control-theory as propounded by the Lubanga majority may be too one-dimensional. What we should be looking for is a more comprehensive model of joint perpetratorship which contains both subjective and objective elements. Typical factors are a person's involvement in the planning of a joint enterprise, his mens rea, and possibly his personal interest in the success of the enterprise; furthermore, the importance of his contribution to the success of the criminal plan and the proximity of his contribution to the actual commission of the offence. It is a matter of debate whether greater emphasis should be placed on subjective or on objective factors. But only if we recognize that there is a cluster of different considerations that may be relevant can we rationally lead that debate.

3.1. Can modes of liability be combined at will?

In our view, Van den Wyngaert was right to express caution about this mode of liability. First and most importantly, Van den Wyngaert argued that there was nothing in the text of Article 25 to justify such ad-hockery.Footnote ⁴⁵ Article 25(3)(a) talks of perpetrating ‘jointly with another’ and ‘through another person’, but does not mention the possibility of combining these modes of liability.Footnote ⁴⁶ Can separate modes of liability be combined without special justification?Footnote ⁴⁷ There are normative reasons to be sceptical of such combinations and it must be determined whether they are consistent with international criminal law in general and the ICC Statute in particular. In the case of indirect co-perpetration, one might argue that the defendants in such cases will independently satisfy the criteria for each mode of liability. Consequently, the demands of both co-perpetration and indirect perpetration are independently satisfied. So what could be the harm in that?

Indirect co-perpetration involves something more than a straightforward application of the concepts of indirect and co-perpetration as those terms are used in Article 25 of the ICC Statute. The Katanga and Ngudjolo Pre-Trial Chamber decision on the Confirmation of Charges concluded that only with the co-operation of both forces were the defendants able to consummate the international crimes.Footnote ⁴⁸ But it was not the case that each of the defendants met the standards for both co-operation and indirect perpetration. Indeed, if that had been the case, the doctrine of indirect co-perpetration would have been superfluous; prosecutors could simply have selected between co-perpetration and indirect perpetration and could have proceeded with one of these doctrines as their theory of the case. For example, in the ICTY Stakić case, one of the first to attempt application of the control-theory, the Trial Chamber, held that Stakić was responsible as a principal to the crime even though it was subordinates of his co-perpetrator who performed the actual killings.Footnote ⁴⁹ It was therefore not the case that Stakić independently met the standards for indirect perpetration, since he was not personally in control of the vertical organization that performed the killings. The question is whether such doctrinal overreach is a natural outgrowth of the judicial application of the doctrine of indirect co-perpetration or whether it is implicit in the doctrine itself.

None of this suggests that an adequate theory of indirect co-perpetration cannot be constructed. However, it cannot be merely assumed, and that theory is certainly not a straightforward application of the bare text of the Statute. Constructing this theory requires a judicial recognition of how the control-theory has evolved in German criminal law doctrine and how far it should be extended, based on an independent examination of basic principles of criminal law Dogmatik. Some theory must explain this new mode of liability as it emerges from the raw materials of Article 25(3)(a). Through forming a common intent, both Katanga and Ngudjolo allegedly reached beyond their own troops. They built a team of two, a collective with power over both of their troops. This fact may justify a legal attribution of the acts of Lubanga's troops to Ngudjolo, and vice versa.Footnote ⁵⁰ However, an adequate theory needs to carefully distinguish between individuals who control distinct organizations but deploy them towards a common cause and individuals who jointly exercise combined authority over a single vertical organization. The latter constitutes a junta model of indirect co-perpetration via a single apparatus of power, while the former represents a joint perpetration through multiple vertical organizations.Footnote ⁵¹ The judicial standards for each flavour of the doctrine might be different, given the different structure of the ‘control’ in these cases. So far the ICC has not explored these differences to a satisfactory degree.

3.2. The role for organizations under Article 25

Van den Wyngaert also expressed anxiety about the growing importance of organizations within the control-theory. Simply put, she argued that Article 25 applies to indirect control over persons committing international crimes, but not indirect control over organizations committing the crimes.Footnote ⁵² This is a complex point and one that the various Pre-Trial Chambers have shown insufficient dedication to addressing. Of course, all organizations, whether criminal or corporate, are composed of natural persons.Footnote ⁵³ Hence Article 25 is applied to the indirect control of persons even where control is exercised by means of an organization. The question is whether it is appropriate for the doctrine to give a special position to ‘organizations’ as a legally significant mediator that stands between the defendant and the relevant physical perpetrators who commit the actual crimes.Footnote ⁵⁴ For Van den Wyngaert, apparently, the concept of the organization is a distraction that finds no support in the Statute; furthermore it necessarily involves situations where individual control over the street-level perpetrators is increasingly attenuated (through bureaucracy) rather than direct. This objection is debatable. Roxin's theory in fact emphasized that bureaucratic control is, in its own way, ‘immediate’ because the organization carries out the orders of the leader as a matter of course.

If the ICC Appeals Chamber wishes to continue using Organisationsherrschaft within the context of indirect co-perpetration, it ought to come up with a deeper theoretical argument that explains why indirect perpetration through an organization is consistent with the language of Article 25 and its reference to persons. One argument might be that the organization is nothing more than a convenient legal shorthand for the control exercised by the defendant over the street-level perpetrators who performed the actus reus of the crime. Under this approach, Organisationsherrschaft would not be a separate mode of liability at all, and therefore its absence from the text of Article 25 would be irrelevant. Rather, it would be classified as one avenue towards reaching the Article 25 standard of indirect perpetration: under that view, the indirect perpetrator indeed commits the offence through another person, namely the immediate actor. The organization provides the ‘through’ element; it connects the indirect perpetrator with the actor.

Another trend that might support the role of organizations within Article 25 is the growing importance of organizations in other areas of the substantive doctrine of international criminal justice. For example, the requirement of an ‘organizational plan or policy’ in crimes against humanity is now an accepted element along with the requirement of a widespread or systematic attack.Footnote ⁵⁵ Although this element is controversial, and some would prefer to eliminate it, the recent cases at the ICC are built around it.Footnote ⁵⁶ For example, the indictments against the Kenyan suspects relied on the existence of two organizations – the Mungiki and what the ICC dubbed the ‘Network’ – to fulfil both the organizational requirement for crimes against humanity and also the organizational requirement for an organized apparatus of power.Footnote ⁵⁷ A coherent judicial theory of ‘organizations’ within international criminal law would have to holistically address all of these issues with one theory of macro-criminality.

3.3. Dolus eventualis

The final – and perhaps most important – objection to the ICC's continued reliance on indirect co-perpetration involves the application of dolus eventualis to these cases. The term dolus eventualis is notorious for competing and often conflicting definitions; the term may very well obscure more than it illuminates. Scholars at the domestic and international levels have argued extensively over whether it includes a so-called ‘volitional’ component, i.e. an attitude by the defendant (either resignation, reconciliation, approval, or consent) regarding the future event. There is also a parallel debate about whether dolus eventualis is similar to – or greater than – its common-law cousin, recklessness. However, all definitions agree that dolus eventualis involves liability for foreseeing the mere possibility of future events.

One way of extending criminal liability to cases where the actor does not directly foresee the harmful result has been devised by the ICTY. Under the third variant of JCE (JCE III), a participant in a JCE is held responsible for any reasonably foreseeable act of any other participant done in furtherance of the joint enterprise. JCE III thus allows convictions – as a principal, no less – for crimes committed by co-venturers that fall outside the scope of the agreed-upon plan. The ICC has so far been reluctant to adopt JCE doctrine. By refusing to read JCE doctrine into Article 25(3)(d) of the ICC Statute, the ICC developed a reputation for analytical rigour with regard to criminal-law theory. Yet, at least part of the rationale for the ICC's adoption of control-theory, though never explicitly stated as such,Footnote ⁵⁸ may have been to reach similar results as would have been possible under JCE without explicitly adopting that much-maligned doctrine.

Indirect co-perpetration, as applied by the ICC, may indeed not be that much different from JCE III and its Pinkerton-like vicarious liability.Footnote ⁵⁹ A structural similarity appears as soon as liability for dolus eventualis is imposed. To be sure, there is conflicting precedent in this area, and some ICC Chambers have concluded that dolus eventualis is not applicable to most crimes prosecuted before the court.Footnote ⁶⁰ But some Chambers have declared that dolus eventualis (or some version of it) is consistent with Article 30; and when combined with indirect co-perpetration, the two produce a result that is very similar to the controversial JCE III.Footnote ⁶¹ Specifically, indirect co-perpetration allows the defendant to be convicted for the crimes committed by the virtual apparatus of power deployed by one of the defendant's co-perpetrators. If dolus eventualis is added to the mix, a conviction is allowed even if the crimes committed by the organized apparatus of power were not part of the criminal endeavour agreed to by the horizontal co-perpetrators. The only limiting principle, according to the Pre-Trial Chamber in Lubanga, is that the defendant must have been aware of the possibility that such crimes might be committed and must have reconciled himself to that possibility or consented to it.Footnote ⁶² Indeed, the Pre-Trial Chamber recognized that ‘if the risk of bringing about the objective elements of the crime is low, the suspect must have clearly or expressly accepted the idea that such objective elements may result from his or her actions or omissions’.Footnote ⁶³ The result is substantially similar to JCE III because both the defendant and his co-perpetrator are prosecuted as principals even though their attitudes to the crime are quite different. The defendant did not desire the crimes but nonetheless realized that they might occur; his co-perpetrator, however, might have indirectly perpetrated the crimes with full-blown intent. Whatever the merits of this theory, one should not pretend that it departs significantly from the underlying premise of JCE III.Footnote ⁶⁴ Both involve liability for risk-taking behaviour.Footnote ⁶⁵

None of this is to suggest that dolus eventualis is not a legitimate mental state, used in many jurisdictions, to ground criminal culpability. It certainly is. Rather, the doctrinal question is whether it is consistent with the default mens rea standard expressed in Article 30 of the Statute, which states that the applicable mental state shall be intent and knowledge ‘unless otherwise provided’. Intent and knowledge require awareness that the consequence will occur ‘in the ordinary course of events’. Most of the authoritative commentators writing about the ICC Statute drafting process agree that dolus eventualis was not explicitly covered by the Article 30 standard. For example, Roger Clark famously wrote that ‘dolus eventualis and its common law cousin, recklessness, suffered banishment by consensus’ at Rome.Footnote ⁶⁶ The key piece of evidence is the draft of the 1996 Preparatory Committee report, which included a subsection on dolus eventualis and recklessness that was subsequently deleted and never made it into the final version of the ICC Statute.Footnote ⁶⁷

To reiterate the point, none of this demands a form of global scepticism regarding dolus eventualis. Rather, the point is simply that its application to situations of indirect co-perpetration is powerful, and the ICC must tread carefully to determine whether it is consistent with the principle of individual culpability. A defendant charged with being an indirect co-perpetrator is held responsible for actions perpetrated by a vertical organization that he does not personally direct. In cases of dolus eventualis, the crimes are not desired by the defendant but merely foreseen as a potential risk. What is this hypothetical defendant's level of culpability? Surely such individuals are guilty of some offence under some mode of liability, but that is not the question here – the issue is whether they are deserving of Roxin's label of Täter hinter dem Täter, the mastermind who stands behind the criminal operation. Given that indirect co-perpetration is not explicitly listed in Article 25, the ICC ought to be certain that such a classification is warranted. As it stands now, none of the previous opinions of the court have yet accomplished this task.

4.1. Normative approaches to criminal participation

According to the proponents of control-theory, Article 25(3)(a) is an expression of what can be called a normative approach to criminal participation: the principal is the one who is ‘most responsible’ in the sense that he or she has decisive influence on the commission of the crime, without necessarily physically committing it. This contrasts to what can be termed the naturalistic or empirical approach to liability, which takes as a starting point the natural world and the reality of cause and effect.Footnote ⁶⁸ In ‘empirical terms’, the perpetrator is the one who performs the material elements of the offence and thus ‘perpetrates’ or ‘commits’ the crime. The accessory or accomplice is the one who contributes to causing the actus reus. Anglo-American complicity law, based on a physical concept of ‘commission’, is the classic example of the empirical approach to criminal participation.

The empirical system can be referred to as a bottom-up system. If it is applied to a complex structure of criminal co-operation, say an army, one starts with the soldier who killed a civilian, and then moves on to his superior who gave the orders, and then further on to the government minister who devised the relevant policy. The normative approach, on the other hand, represents a top-down system. One starts with the person who has the main responsibility, for example the minister, and then works one's way down to the smaller fry in the lower echelons of the military unit. Thus, in the Anglo-American complicity scheme the government minister is an accessory, while on the basis of a normative system like the control-theory he is a principal.

The ICTY has adopted a normative approach with regard to JCE liability. The Appeals Chamber in Tadić referred to JCE, or common purpose as it was then termed, as ‘a form of accomplice liability’.Footnote ⁶⁹ But it also used the term ‘perpetrator’ and ‘co-perpetrator’ to refer to a ‘participant’ in a JCE.Footnote ⁷⁰ Moreover, it brought JCE liability under the heading of ‘committing’ and distinguished it from aiding and abetting, which, it felt, understates the degree of criminal responsibility.Footnote ⁷¹ These findings sparked a debate amongst Trial Chambers in subsequent cases. There were judges who felt it important to adhere to the principal/committing–accomplice/participation distinction,Footnote ⁷² and there were others who thought that the principal–accomplice classification is immaterial because it is at sentencing level that variance in roles is expressed.

Without going into details of the debate,Footnote ⁷³ it suffices to conclude that with JCE a categorization of offenders along normative lines was introduced to ICTY case law. In the Odjanić Decision, the majority of the Appeals Chamber – over a strong dissent by Judge Hunt – affirmed that ‘joint criminal enterprise is to be regarded, not as a form of accomplice liability, but as a form of commission’.Footnote ⁷⁴ With the latter position, aiding and abetting developed into a mode of liability that is considered less blameworthy than participation in a JCE. This was confirmed in Šljivančanin, where the Appeals Chamber held that ‘aiding and abetting is a lower form of liability than ordering, committing, or participating in a joint criminal enterprise and may as such attract a lesser sentence’.Footnote ⁷⁵ Research into international sentencing confirms that aiding and abetting has been treated, by the ad hoc tribunals, as less blameworthy than other modes of liability.Footnote ⁷⁶ By now it is safe to assume that a mitigation principle applies at the ICTY with regard to aiding/abetting versus JCE liability, which shows the influence of a normative approach to criminal participation.Footnote ⁷⁷

The same hierarchy of blameworthiness with regard to aiding/abetting vis-à-vis JCE can be found in SCSL law. The conviction of Charles Taylor is interesting in that respect. Taylor was convicted for aiding and abetting war crimes and crimes against humanity and sentenced to 50 years in prison.Footnote ⁷⁸ Had he been convicted for JCE, the sentence would have been higher.Footnote ⁷⁹ Still, 50 years is a serious sentence in the overall SCSL sentencing practice. While aiding and abetting is considered a lesser form of liability, it does not automatically imply a lenient sentence; leniency is a relative concept.Footnote ⁸⁰

Preference for the normative model is even stronger at the ICC. The majority of the Court strictly adheres to the distinction – introduced by the control-theory – between principals and accessories and distinguishes between principal liability (‘committing’) in subparagraph 3(a) and accessorial liability (‘contributing to’) in subparagraphs (3)(b–d) using a normative approach.Footnote ⁸¹ Charging defendants as intellectual or remote principals under 25(3)(a) means that they played a central role, that they had ‘control of the crime’.Footnote ⁸² This is contrasted to liability under subparagraphs 25(3)(b–d) where control plays no role and accessories are regarded as less responsible and less blameworthy.

But also at the ICC the normative approach of the control-theory has been subject to dispute. Judge Tarfusser, in his dissent to the Appeals Decision in the Regulation 55 case in Katanga, referred to the Court's case law on Article 25(3) as ‘far from . . . uncontentious or settled’;Footnote ⁸³ and Judges Fulford and Van den Wyngaert in their concurring opinions manifestly disagreed with the normative hierarchy that the control-theory (allegedly) creates in Article 25(3) of the ICC Statute.

From the viewpoint of comparative law, it is interesting to note that the majority opinions in Lubanga (PTC and TC), on the one hand, and the concurring opinions of Judges Fulford and Van den Wyngaert, on the other hand, reflect a clash of legal cultures that, interestingly, cuts across the civil-common-law divide. The majority views stand for a ‘dogmatic’ concept of perpetratorship and accessorial liability, which expects the substantive law to reflect (or describe) subtle differences in the measure of responsibility and seeks to establish criteria that permit distinctions between forms of liability (and, consequently, degrees of blameworthiness). This approach is typical for the German/Hispanic tradition. The minority approach, by contrast, looks at the substantive criminal law more from the perspective of the legality principle: it is enough that the definitions of the general and special parts capture as comprehensively as possible all potential forms of reprehensible conduct. Differentiating between degrees of responsibility is not the purpose of offense descriptions or of rules of the general part, but is left for judicial sentencing. This reflects the French and the Anglo-American traditions.

Leaving these comparative-law considerations aside, in what follows we try to shed some light on the reasons and implications of the dispute between the majority opinions and the Fulford and Van den Wyngaert opinions.

4.2. Accessorial and derivative liability

Problematic in the control-theory is the use of the term ‘accessorial’ and the normative meaning attached to it. Consider the Katanga and Ngudjolo Confirmation Decision with regard to indirect co-perpetration viz-a-vis ordering:

‘Accessorial liability’ is equated to ‘lesser liability’. This is confusing to the extent that it does not comport with the empirical model of criminal participation where ‘accessorial’, as non-principal/accomplice liability, has no normative connotation.Footnote ⁸⁵ It merely indicates that liability is ‘derivative’, i.e. that liability depends on the principal crime. Indeed, the modes of liability in Article 25(3)(b–d), by requiring the crime to be at least attempted, and by criminalizing a contribution to a crime, constitute forms of derivative liability. They differ from principal liability in subparagraph 25(3)(a), where ‘commission’ of a crime is defined. Clearly Judge Van den Wyngaert reasoned from an empirical approach to criminal participation when she argued that the principal–accessory distinction in Article 25(3) of the ICC Statute is merely ‘conceptual’ and that it should not translate to a ‘different legal treatment’.Footnote ⁸⁶ We agree with that position to the extent that the terminology in Article 25(3) (b–d) cannot in itself be taken as a normative indication of lesser blameworthiness. Support for this position can be taken from comparative criminal law; there is no rule or theory that categorically links accessorial/non-principal liability to lesser responsibility.Footnote ⁸⁷

Having said that, aiding and abetting in subparagraph (3)(c) can be regarded as less blameworthy vis-à-vis joint perpetration in subparagraph (3)(a). Where accomplices such as aiders and abettors are further removed from the centre of the commission of the offence, their responsibility is reduced in comparison to perpetrators, and consequently their punishment should be reduced as well. This interpretation of aiding and abetting comports with the previously mentioned normative approach to aiding and abetting in international jurisprudence.

In our view, the hierarchical distinction among four types of perpetrators and accomplices, as proposed by the Lubanga Pre-Trial and Trial Chamber and the Pre-Trial Chamber in Katanga and Ngudjolo, does not adequately reflect the normative relationship between these types of participant. Ordering, soliciting and inducing others to commit crimes in subparagraph 3(b) is not necessarily less blameworthy than indirectly perpetrating a crime as penalized in subparagraph (3)(a). Even Kai Ambos, a staunch supporter of the control-theory and its (alleged) hierarchical structure,Footnote ⁸⁸ admits that the hierarchy in Article 25(3) is ‘less evident with regard to subparagraph (b) – especially with regard to ‘ordering’ which belongs structurally and systematically to subparagraph (a)’.Footnote ⁸⁹

Moreover, it is highly questionable whether Article 25(3) is based on a single coherent, normative theory of participation. Nothing in the drafting history of the ICC suggests that Article 25(3) was to constitute a self-contained system of criminal participation with a coherent doctrinal grounding. To the contrary, as the chairman of the Working Group on General Principles recalls, drafting Article 25(3) posed great difficulties to negotiate; eventually, a near-consensus was reached where there would be one provision to cover the responsibility of principals and all other modes of participation. Article 25(3) was to provide the Court with a range of modalities from which to choose.Footnote ⁹⁰

Bearing in mind this ‘aggregate’ background of Article 25(3) of the ICC Statute, it seems that Article 25(3) reflects elements of the normative and empirical approaches to criminal participation without clearly distinguishing between the two. The fact that Article 25(3)(a) provides for instances of intellectual perpetration does not make it the sole theoretical grounding for the whole of Article 25. Nor does it relegate all participants covered by subparagraphs (b–d) to lesser liability. The ICC, in forging its path to identifying the responsibility of principals versus accessories, has been too rigorous in drawing lines according to the vague legislative concepts embodied in Article 25(3) of the ICC Statute.