1 Introduction

Customary international law (CIL) is particularly vulnerable to the accusation that it is no more than ‘mere assertion’, a creation of the courts, if not downright fantasy.Footnote ¹ Yet it is in CIL that one finds the strongest claim to objectivity in international law. It is expressed in the doctrine that one of the elements of CIL is state practice, which represents the ‘objective’ element of CIL. It is thought to supplement the ‘subjective’ or ‘psychological’ element of CIL: opinio juris.Footnote ² Elsewhere, I have analysed opinio juris and concluded that it is much less ‘subjective’ than is commonly assumed.Footnote ³ Here, I will argue that state practice is much less ‘objective’ than is commonly assumed.

I will argue that the notion of state practice as a set of ‘material facts’ that should be ‘identified’ and from which customary norms can be ‘induced’ is grounded in obsolete epistemology. The identification of state practice is more adequately described as a selection of what deserves to be counted as state practice. I will argue that the starting point for this selective process is opinio juris. Opinio juris does not come after the fact, as a subjective feeling of obligation that is superadded to a set of otherwise objective facts. Opinio juris is the indispensable conceptual framework without which habits and usages cannot even be ‘seen’ as state practice.

2 Gold-Digging

It is unclear what the adjectives ‘objective’ and ‘subjective’ refer to. To the elements themselves or to the ways in which these elements are studied? The latter option is not very plausible, for why may the convictions of states (opinio juris) not be dealt with in an objective manner? Any sociologist can tell you that it is possible to examine convictions and beliefs. And even if that would not be the case, why would one officially announce that half of the process of CIL identification is carried out in a ‘subjective’ manner?

No, apparently these adjectives refer to the elements themselves and not to the investigator. But then the question arises, why is state practice labelled an ‘objective element’? In what sense can we say that acts, words, conventions and usages are objective elements? Can we speak at all about ‘objective elements’? Are molecules and stars objective elements? And again, what is so subjective in opinio juris if we understand that notion as the conviction that a certain norm is a legal one? The more one thinks about it, the stranger it is.

The only way to make sense of the use of these terms is to take into account another dichotomous pair of words by means of which the two elements are distinguished: the ‘normative’ and the ‘factual’. Apparently, the claim is that whereas opinio juris is the ‘normative’ element, state practice consists of ‘facts’. There are three interrelated assumptions at stake here.

1. (1) The nature of the object that is investigated: facts.

   • Most authors understand practice as a heap of ‘material facts’. State practice is called ‘a sort of … raw material for custom’,Footnote ⁴ an ‘inert mass of accumulated usage’Footnote ⁵ or – in a modern version – a set of ‘raw data’.Footnote ⁶

2. (2) The method to be followed: induction.

   • The account of state practice as a heap of social facts gives rise to the idea that the analysis of state practice can be conducted as any inquiry into ‘facts’. They can be collected as ‘data’, which should be ‘described’ and which forms the starting point of an inductive process of reasoning that moves from the described facts upwards to general statements about the rules that are followed.

3. (3) The aim of the investigation: identification.

   • The assumption is that state practice should be investigated in such a way that possible candidates for CIL can be ‘identified’. The term ‘identification’ suggests that CIL is thus found rather than constructed.

The combination of these three related claims can be captured by the metaphor of gold-digging: the gold of CIL can be found by digging the inert material mass of sand and clay, and by bringing it upwards to the surface, after which it can be identified by sieving the sand.

Several authors have criticised this picture by criticising assumption (2). They think that an inductive description of the facts is only half of the work to be done. Roberts,Footnote ⁷ for instance, thinks that the ‘descriptive accuracy’ of an investigation of state practice should be complemented by a deductive method of searching for ‘substantive normativity’, starting with normative ideals and then descending to the formulation of rules. In a similar vein, TalmonFootnote ⁸ believes that we should make more room for deduction, especially where state practice is inconclusive, non-existent, contradictory or inconsistent with opinio juris. Finally, MerkourisFootnote ⁹ wants to complement the inductive method of rule identification with a deductive method of rule interpretation.

Although these authors have different objectives in mind, they are united in their view that describing practices is not enough to account for CIL, and that inductivism should be complemented by deductivism. They rightly point out that the problem of a merely inductivist investigation of state practice is both unrealistic and impossible, but they seek the solution of that problem by adding a normative or deductive approach. And through this search for an additional approach they leave intact the doctrinal picture of an initial investigation of state practice as a purely descriptive and inductive enterprise. Inductive ‘identification’ and deductive ‘interpretation’ are then seen as separate activities. Assumption (1) (practice as a collection of facts) as well as assumption (3) (identification of rules as the main aim of any investigation of state practice) remain untouched.

I will argue, however, that not only the assumption of induction (2), but the entire picture of gold-digging, is largely illusionary. I will start by addressing the assumption of induction (2) and take advantage of well-known insights that were developed in the philosophy of science. Consequently, I will criticise assumption (3), according to which identification is prior to interpretation. By using the insights of hermeneutics, I will argue that opinio juris is the starting point for any investigation of state practice and helps to define both state practice and CIL as a whole.

3 Loaded Perceptions: The Problem of InductivismFootnote ¹⁰

Let us put ourselves in the position of an international courtFootnote ¹¹ which is confronted with the task of deciding a case or of delivering an advisory opinion by distilling or identifying rules from a raw heap of facts. Such a court has to carry out two types of translation. In the first place, it should proceed from particular instances to general statements that deal with categories; in the second place, it has to proceed from facts to norms. The activity of such a court can be understood as similar to the interpretative activities of a domestic court which is asked to apply a given statutory rule to a concrete fact situation. Both courts are confronted with the task of translation; but they walk in opposite directions. Whereas an international court moves from particular fact situations to general norms, the domestic court moves in the opposite direction by applying general norms to concrete fact situations.

The latter process of interpretation is beset by difficulties that are abundantly analysed in the literature on judicial interpretation and decision-making. But in one aspect these tasks are easier to accomplish than the inductive translation from particular facts into general norms, and that is that in interpretation one knows where to start. It is with a set of possible rules – general and normative statements – in mind that the facts are investigated. Did the defendant buy his knife before or not? Was the will drawn up in the presence of a notary or not? Rules guide the investigation and select the facts or the sets of facts that are relevant in the light of the rule. The rule highlights that part of factual reality which is possibly relevant. This is not to say that those parts that are not highlighted remain in obscurity forever. The outcome of interpretation may be downright undesirable and may provide reason to re-open the investigation and to search for other rules which may highlight other aspects which become relevant and which may generate a different outcome. This is what Llewellyn means when he talks about the necessary creativity of judges to erect alternative ladders through the legal material in order to arrive at the desired outcome. But alternative ladders also start with a rule that guides the investigation.Footnote ¹²

Such starting points are absent if we consistently think of an inductive investigation of state practice. Then, it is supposed, we start with the facts and nothing but the facts. But which facts? There are a multitude of facts, events, gestures, movements, usages, conventions, declarations and resolutions, and we do not know how to make sense of such facts without any preconception in mind about what we hope or expect to find.Footnote ¹³

This is obvious in even the most basic form of perception. As Hanson eloquently pointed out as early as 1958,Footnote ¹⁴ seeing involves recognising patterns. We can of course see black and white pixels and we can be sure everybody sees the same pixels. But as the well-known picture of rabbits and ducks illustrates, we can see these pixels in different ways, ordered in different patterns. And these ordering patterns are to a large extent informed by what we expect to see in a particular context as well as by our background knowledge. If I stare into the microscope, I might see beautiful colours but I do not really know what I see and I therefore see nothing at all. There is no simple observation, therefore, of ‘what is there’. Our perception is loaded with preconceptions, conceptions, expectations and patterns that all lurk in the background. They should not be shunned and avoided, for if we were to succeed in discarding them, we would no longer know what we see. Or, as Heidegger succinctly put it: ‘Only those who understand can hear.’Footnote ¹⁵

If it is already difficult to make sense of dots and lines without any context that guides expectation and makes meaningful observation possible, how would we be able to understand things as elusive as ‘acts’, ‘usages’ and ‘conventions’? I am not referring here to Hart’s emphasis on an internal point of view. My claim is that even researchers who take an external point of view and regard people’s doings like the movements of ants just have to start from some theory in order even to study such ants. As any sociologist or anthropologist knows, without a basic theoretical framework it would be impossible to select what kinds of phenomena are worth observing. If we add to this that international lawyers are supposed to investigate not only ‘practice’ (a complex phenomenon in itself) but ‘state’ practice, which is a legal notion, it is clear that the facts which are observed cannot possibly be considered as ‘raw data’ or ‘raw material’.

4 Law as Theory and Object

Fortunately, that is not what legal scholars or courts do. Just like natural and social scientists, they start with a theoretical hypothesis that provides for the patterns that order the sense data in a meaningful way. Physics conceptualises light as either ‘wave’ or ‘particle’ and hence we ‘see’ light as waves or particles. Psychological concepts such as ‘hysteria’ and ‘burn-out’ are not merely different terms for the same phenomena but constitute different phenomena because they order facts in a different way.

So, too, courts and legal scholars start from a background theory. But there are three important differences between the background theories of social scientists and those of legal experts.

1. (1) Law as theoretical framework

   As I have argued elsewhere,Footnote ¹⁶ legal research also starts from a theoretical framework, but, unlike social scientists who develop a theoretical framework that is independent from the object they investigate, the framework for legal research is the law itself. Legal notions and conceptions form the framework and pattern by means of which facts are ordered. Not only terms such as ‘seabed’ but also ‘sovereignty’ or ‘specially affected’ are notions that guide the investigation and furnish the patterns that enable us to see a point in the ‘raw data’. Whereas in the social sciences there is a certain distance between theory and object, the peculiar feature of legal scholarship is that its theoretical framework and its object are identical. Of course, this does not apply to empirical legal studies or, for that matter, to philosophy of law, where the theoretical framework may (and should) be broader, but most doctrinal legal research revolves around the question of how certain novel social or technical developments or problems can be regulated such that they can be fitted into the legal system or in a way that is coherent with underlying principles, or, alternatively, how different legal arrangements from different legal orders can be integrated and harmonised. In all these studies, law is not only the object of research but also provides the theoretical framework consisting of legal concepts, standards and exemplars.

2. (2) Definitional concepts

   The fact that the function of a theory is taken over by the law itself and that therefore the facts are examined by means of legal concepts entails another peculiarity, which has to do with the nature of legal concepts. Legal concepts are not descriptive but definitional. In law, such definitions are usually couched in terms of conditions, although one also encounters enumerative definitions.Footnote ¹⁷ When we say that some paper p is a ‘contract’, we say that this paper p meets the conditions that should be fulfilled in order to count as a contract. Legal general statements are of the form: For all X, if conditions a, b and c apply, they count as Y.

   This statement does not generate new knowledge but is a definition of Y.

   If we start investigating state practices by means of such definitional concepts and apply the filter of the law to the facts, we examine that factual reality by enquiring into whether reality can be categorised into these definitions. We differentiate between items that do and items that do not, or only partially, fulfil the conditions that are contained in the general statement.

   This strategy is applied at all levels. At the micro-level, we start for instance by investigating whether a certain contract x meets the conditions for ‘contract’ and can be considered a valid one; at the meso-level, we might ask whether conditions for ‘sovereignty’ are met; and at the most abstract level, we inquire into the degree to which a certain usage counts as ‘practice’. Definitional concepts are very useful in a deductive process in which particular instances are examined for their conformity to the conditions and thus ‘put to the test’, but they are emphatically not the result of generalisations from particular facts. They are the result of our decision or of convention that defines a concept in terms of conditions. Their existence cannot be tested by reference to empirical facts but by referring to these conditions. That is why Kelsen could say that ‘the existence of a legal norm is its validity’.Footnote ¹⁸ Contracts, sovereign states and practices exist insofar as they meet the conditions.

3. (3) No falsification by facts

   This difference between descriptive and definitional concepts implies a third difference as well. Whereas scientific general statements can be refuted or falsified by recalcitrant facts, that is not the case in legal research. Scientific theory does not merely contain general statements such as ‘if light has a wavelength of 400 nm we call it “violet”’. If that were the case, science would be just like legal scholarship and would proceed by definitional concepts based on convention. But although science makes use of conventions, and although observations on colours cannot be conducted without such conventions on numerical values, its general statements express new – hypothesised – correlations: hypotheses like ‘violet flowers attract more butterflies than red ones’, which can be falsified when a species of butterflies is discovered preferring red flowers. In legal research, it is not the theory that is falsified but the facts that are proven to be deficient. If we encounter papers that fulfil only half of the conditions for contracts, we do not think that our general statement is falsified; we just say that those papers are not ‘proper’ or valid contracts.

Is there then no sense in the idea that rules can be inferred by means of induction? Could we not say, ‘Here is a contract which is signed and there is a contract which is not signed, so maybe signature is not a decisive condition’?

That can happen indeed. In international law (IL), the conditions for validity are often not clearly defined, are ambiguous or contested, and indeed it is possible to adjust our set of conditions somewhat to match reality, but that can be done only in the presence of other features that qualify a certain item as a suitable candidate for being examined as a potential ‘contract’. If none of the conventional conditions are fulfilled, we do not even start investigating whether a paper counts as a contract or whether we should adjust our notion of contract in light of such divergent practices. We may for instance investigate the practices around marriage and see how in different cultures marriages are conducted and eventually end up with a wider concept of marriage than we initially had in mind. But such practices at least should have a credible claim to be considered as ‘marriages’. If they do not even remotely look like marriages, we would not take the trouble to contemplate them as serious candidates for being considered as marriage.

This also applies to concepts such as ‘sexual slavery’. They do not have a claim to validity, of course, but it should be possible to claim or suggest that the term might be appropriate to the case at hand. Further investigation may lead to limitation or expansion of the concept (and may even be extended to such an extent that it borders on ‘marriage’), but again there should be some shared features in order to conduct such an investigation in the first place.

To say that rules cannot be inferred from facts is not the same as saying that rules cannot emerge out of facts. I think that this is very well possible and discernible in all those instances in which patterns of behaviour slowly develop into rules.Footnote ¹⁹ But we should make a difference between ontological views on how rules are formed and epistemological views on how they can be identified.Footnote ²⁰

In investigating state practice, legal decision-makers necessarily see the practices, usages and customs through the filter of rules and definitional concepts that they already have in mind. It is not necessary that these rules have an unambiguous status or are already legally valid. As Hakimi remarked, it is more to the point to refer to them as ‘normative positions which may not be fully supported’.Footnote ²¹ But whether they are formally valid, endorsed by a majority, soft law or only half-baked rules in the making, they nevertheless function in the same way as theoretical hypotheses do in the sciences: they guide our perceptions and investigations by selecting the items that are ‘relevant’ and have ‘legal salience’.Footnote ²²

At this point one may wonder, how is it possible to distil and identify rules if all we can do is to investigate the facts by means of … rules? The apparent circularity of opinio juris, much criticised on the ground that it presupposes the existence of law in order to identify law, seems to be repeated also in the examination of state practice! And indeed, circles abound. Any examination of practices or facts presupposes the guidance of rules. But which rules should be selected? In order to decide on the appropriate filter, the facts should be consulted. And how can the facts be consulted? By the guidance of rules. This treadmill is a well-known feature of interpretation and is called the hermeneutical circle.

According to Heidegger, such circles should not be shunned, as they reflect the structure of our existence.Footnote ²³ But even without such metaphysics, it is important to see that it would be a misunderstanding to discard such circles as mistaken methodology or faulty logic. The preconceptions which are necessary to select that which is meaningful may be seen as ‘prejudice’, but it is a kind of prejudice which is essential for any understanding.Footnote ²⁴ It is therefore worthwhile to be open about the inevitability of such circles. Instead of sweeping them under the carpet it is better to acknowledge them and to make clear where such circles arise and which points they cover. If we talk, therefore, about the identification of rules by examining state practice, we should acknowledge that such an examination can be carried out only if we have standards, rules and normative positions in mind as a background theory. This does not imply that we can only discover what we already have in mind. After the initial selection of the relevant material, its weight should still be assessed. But to represent this ‘material’ as a heap of objective facts is a mischaracterisation.

5 Common Law?

Schauer and Bradley suggested that this process is very similar to common law adjudication.Footnote ²⁵ They rightly point out that the examination of state practice is not the kind of inductivist enterprise that is commonly and officially announced. They propose that the identification of rules by an investigation of state practice should be regarded as an attempt to construct chains of precedents. Schauer refers to the fact that different chains of precedents can be constructed, just as Llewellyn pointed out that different ladders can be built by means of which it is possible to motivate different outcomes.

At first sight, one may question the wisdom of the term ‘precedents’. State practice comprises not only precedents but acts, usages and conventions of all kinds. It is only in the eyes of a judicial decision-maker that such acts can be labelled as ‘precedents’: events that preceded a new case and might have a bearing on the ways in which a new case should be resolved. But this is not necessarily a problem. Both authors start from the perspective of the adjudicator and in such a perspective state practice is examined in the light of the theoretical framework of the law. The clear advantage of this comparison with common law adjudication is therefore that it openly concedes that the adjudicator runs in the hermeneutical circle. It makes clear that the examination of state practice is not just an investigation of raw material facts but presupposes a selective filter of possibly relevant legal perspectives, from which acts and usages are indeed regarded as ‘precedents’ for the case at hand. The plausibility of the perspectives is judged by means of a further investigation of the facts, which necessarily presupposes a theoretical background perspective, etc.

However, it seems to me that this account, although much more realistic than official objectivist inductivism, assumes too much: it presupposes that it is clear what counts as precedent. In domestic common law adjudication that may be indeed the case, at least to a certain degree. The legal material is to some extent clearly defined, although here, too, the weight of the precedents may vary. Yet, it is within the domain of this legal material that the interpreter can choose to construct different chains of precedents leading to the decision in the new case.

In CIL, the boundaries of the legal material are less clearly demarcated. The questions to be answered are therefore not only how to construct different chains of precedents. Before those questions can arise, a prior question must be answered and that is whether a certain usage may count as precedent: is a certain act, text, omission weighty enough to count? This is clear for instance from the wording of the 1969 North Continental Shelf case, in which it is argued that despite the fact that the principle of equidistance is used by the majority of states and in numerous cases, there are ‘several grounds which deprive them of weight as precedents in the present context’.Footnote ²⁶ State practice is not examined in order to decide on the applicable chain of precedents and cannot merely be regarded as a matter of choice between various applicable chains. The court that examines state practice does so in order to find out which instances might qualify to be labelled ‘precedents’. Only after that can their respective weight be ascertained and the connecting chains constructed.

6 The Double Function of Opinio Juris

Surprisingly, this prior question is usually answered by referring to opinio juris. In the Asylum case between Colombia and Peru (1950), for instance, the ICJ asserts that all the instances of granting asylum, mentioned by Colombia, simply do not count as precedents:

Institutions that arose in an extra-legal context lack legal relevance and do not count as ‘precedents’. In the Nicaragua case, too, state practice is examined by reference to opinio juris, where it is argued that intervention is not undertaken with a sense of a legal right:

The reader might observe that these passages show exactly what is expressed by the official doctrine, which requires the presence of both an objective and a subjective element. Opinio juris is explicitly invoked.Footnote ²⁹

I would, however, say that the influence of opinio juris is much stronger than is officially conceded. The official claim is that there is first an – independent – examination of state practice, and that that practice can only be regarded as CIL if it is accompanied by opinio juris. What we see here, however, is that without opinio juris a recurrent and widespread usage simply does not count as legally relevant state practice. It counts as mere ‘convention’, as comity, etiquette or rules that are followed merely with a view to ‘expediency’ or ‘international policy’. Opinio juris is required in order to ‘see’ something as relevant practice instead of merely a heap of facts or usages.

Opinio juris, therefore, indeed ‘counts twice’:

1. (1) A usage/act counts as ‘practice’ if carried out with a sense of legal obligation (opinio juris).

2. (2) A practice counts as an element of CIL if carried out with a sense of legal obligation (opinio juris).

Opinio juris is a condition not only for ‘identifying’ CIL, but also for differentiating those usages that might be seen as legally relevant practice (‘precedents’) and those which are not.Footnote ³⁰

This second function is usually not recognised in the official accounts of state practice as an objective element but it is in line with what I remarked on above about the necessity of a theory. In order to ‘see’ something as practice, the law functions as the indispensable theoretical searchlight – a searchlight which is only deemed fit if it has passed the test of opinio juris! That is why torture or pollution, although they are clearly practices that are widespread, uniform and consistent, do not ‘count’ as practices. That is why granting asylum can be regarded as legally irrelevant and as mere friendliness towards neighbours, as is maintained in the Asylum case. The objective element is therefore not only connected to the subjective element:Footnote ³¹ state practice cannot even be investigated without opinio juris. Opinio juris is the starting point for any meaningful investigation of state practice.

Of course, this is not to say that state practices cannot be normatively meaningful without opinio juris. As I noted above, we should be careful to distinguish the (ontological) formation of normative rules as emerging in and from practices from the (epistemological) identification of such rules. I am now just referring to how the courts in their investigation of state practice identify legally meaningful and relevant normative positions that might qualify as precedents in a certain case. And in order to do so, the opinio juris plays a major role as a theoretical searchlight.

7 A Practice of Claims

It seems then that my reading leads to an even more incongruous result than official doctrine. There, opinio juris is already troublesome because of its circularity. As is pointed out repeatedly, opinio juris identifies a rule of CIL by reference to the conviction of states that they are under an obligation to comply with that same rule. My contention seems to be even more bewildering than this: only those acts, usages and conventions that are carried out with a sense of legal obligation (opinio juris 1) count as state practice, as a set of potentially meaningful precedents, which can only give rise to CIL if coupled with opinio juris (2). There is not one circle; there are two – or more specifically, there is a spiral.

However, as I have pointed out elsewhere,Footnote ³² the circularity is not so devastating if we understand opinio juris not as a conviction or a belief that one is under a legal obligation but as an articulated and publicly accessible claim.Footnote ³³ These claims are usually couched in general and descriptive terms: ‘this is the rule/law/principle we maintain and cherish’. Because of this particular descriptive form we might easily mistake them for descriptions of facts (we are under this legal obligation) and then circularity is indeed vicious. But in fact they are declarative statements: statements that constitute the law by means of a declarative form.Footnote ³⁴ They are not expressions of legality but claims to legality.Footnote ³⁵

If we understand that state practice can only be – and is only – investigated by means of such claims to legality that form the searchlight for the investigation of practices, a number of confusions can be clarified. In the first place, it has already been remarked that a lot of the ICJ’s judgements refer, not to state practice as it is traditionally defined,Footnote ³⁶ but to resolutions, declarations and the like. In a fascinating article Choi and GulatiFootnote ³⁷ presented statistics of the kind of evidence found by the ICJ under the banner of ‘state practice’. They found that acts and practices were hardly investigated at all, and the official sources of state practice (diplomatic correspondence, domestic legislation, etc.) in only a small minority of cases. Instead, in an overwhelming majority of cases treaties and other ‘aspirational and forward-looking documents’ were cited. The authors find that amazing: states enter into treaties in the absence of CIL and it would therefore be strange to cite treaties as proof of the CIL.

Choi and Gulati confirm empirically the uneasiness expressed by Mendelson more than two decades ago, who observed, while commenting on the Nicaragua case (Merits):

And he adds to this:

And indeed it is. This is exactly the case. To lift oneself up by one’s own bootstraps accurately describes the hermeneutic circle.

8 Interpretation Prior to Identification and Application

The observation that the identification of CIL is circular is hardly novel. Koskenniemmi had already observed that ‘doctrine about customary law is indeterminate because circular. It assumes behaviour to be evidence of the opinio juris and the latter to be evidence of which behaviour is relevant as custom.’Footnote ³⁹ However, as we have seen, ‘behaviour’ and ‘evidence’ are not as unproblematic as even Koskenniemi assumes. As Bodansky correctly asserts,Footnote ⁴⁰ if it were only behaviour that is investigated, sociologists would be much better equipped. Legal investigation revolves around texts: an abundance of claims to legality such as resolutions, conventions and treaties. They form the object of investigation.

And they figure at the two levels that we distinguished in Section 7. First of all, they are investigated in order to determine the legally relevant precedents, and secondly they are investigated in order to choose between potentially applicable precedents in order to apply them to a particular case. Both activities are usually conducted in one big sweep. In the judgments of the courts it is just pointed out which texts are considered legally relevant, weighty or convincing. They may be labelled differently, as either ‘state practice’ or ‘opinio juris’, but they are all just texts the relevance of which is assessed and weighed.Footnote ⁴¹

It is not possible to weigh precedents and to apply rules without prior interpretation. That interpretation can be grammatical or historical, but most importantly it is teleological. The texts should be interpreted as having a point. This is noticeable both in the attempt to determine the precedents and in the way in which chains of such precedents are constructed. To begin with the latter: it is obvious that in order to make a choice between different chains of precedents, as in common law adjudication, they should first be constructed as a chain. That can be done by discovering analogies or some aspects these precedents have in common. But this discovery of shared aspects can be conducted only by constructing an underlying principle or rationale. There are no shared features between ‘books’, ‘electricity’ and ‘personal data’. We can only construct their shared features as ‘goods’ to which property law can be applied. And usually this construction is carried out by hypothesising an underlying aim or rationale. Do we follow all those solemn prohibitions of weapons and conclude by analogy that nuclear weapons should be prohibited as well? Or do we follow all – equally solemn – declarations of sovereignty and conclude that such a prohibition of nuclear weapons is not part of CIL?

The same applies, however, also to the first round in which different official acts and texts are examined in order to establish the legally relevant precedents. Here, too, there are numerous candidates, and they can be linked together in different chains, each focusing on different shared features and different underlying principles. Those texts, rules and claims which are not constructed as precedents are denied legal relevance and figure as ‘technical rules’ or as rules of etiquette, comity and friendliness. These ‘legally irrelevant’ rules are obviously not found but interpreted as lacking legal relevance, and they are interpreted by reference to texts that embody claims to legality. Postema adequately described the process of such interpretation of practices and compared this to discerning a pattern: ‘the pattern is likely to be salient because it is meaningful within the practice, rather than meaningful because it is salient’.Footnote ⁴² The degree to which customs can be integrated in such a pattern is therefore decisive, according to Postema.

Although I distinguished for analytical reasons between the two rounds of legal interpretation in which legally relevant material is (a) determined and (b) applied, we should not think of determination and application as sequential in time. As noted above, they are usually carried out in one big sweep: both are the result of interpretation in the light of opinio juris, and they are intricately connected. The more successful the attempt in the first round to deny legal relevance to normative positions and, consequently, the narrower the selection of legally relevant material, the easier the task to choose a preferred ladder in the second round of application. Or, to put it differently, those normative positions that fail to be elevated to the status of ‘legally relevant’ state practice do not need to be examined as candidates for CIL. That is why the double-counting of opinio juris does not come to the surface. Opinio juris is usually already applied in the first round of determining relevant precedents and need not be re-emphasised in the determination of CIL.

It is important to see that both the determination of relevant precedents and the choice between rival chains of precedents presuppose interpretation of the normative material as material that is patterned around its point or telos. The degree to which a claim can be fitted into a pattern is to a large extent dependent on how we construct that telos. As Heidegger in his more lucid moments pointed out, it is not possible to identify something as something (‘Etwas als etwas’) without understanding it as something ‘in order to’ (‘Um-zu’).Footnote ⁴³ Precedents, and also candidate-precedents, are constructed as parts of a Dworkinian chain-novel,Footnote ⁴⁴ or at least – more modestly – as threads weaving a plot in a story.

We should not think of this telos as a set of intentions, the honesty of which can be examined on the basis of diplomatic correspondence between officials. Interpretation is different from grasping intentions at face value. It may very well be the case that the states which signed a treaty had no serious intention at all of furthering the aim of the treaty. They might simply have thought about the reputational costs of making reservations to that treaty, or they were mistaken in thinking that they would not be burdened too much, or had already seen gaps in the treaty that would allow them an exit. If we interpret such texts in terms of intentions, many of those texts can indeed be disregarded as specimens of ‘cheap talk’.Footnote ⁴⁵

However, as Ricœur pointed out, a text (any text, whether novel or academic publication or treaty) gains a certain distance from its drafter or writer. It is this distance that harbours the possibility for the interpreter to discern and to reconstruct several meanings of such texts. According to Paul Ricoeur, ‘what must be interpreted in a text is a proposed world which I could inhabit and wherein I could project one of my ownmost possibilities’.Footnote ⁴⁶

This does not merely apply to my interpretation of literary works; it also relates to how legal texts can be understood.Footnote ⁴⁷ Interpretation is an activity to understand not only what exists but also what is possible.Footnote ⁴⁸ In fact, legal texts – as well as communiqués – are also written with an eye to such possible readings. The importance of textual subtleties can only be understood on the basis of this distantiation. If it were a matter of just unravelling intentions, such care for texts would be unnecessary. But they are highly important in view of the fact that texts live a life of their own. Even if such texts do not reflect the serious intentions of the signatories and are indeed no more than ‘cheap talk’, they can be read as elements of a purposive pattern, that can be constructed as meaningful by the interpreter.

9 Conclusion: Possible Worlds

If we take seriously the notion of interpretation as delineating possible meanings, that means that different chains and patterns are conceivable that order the textual material in different ways. Depending on the choice of such patterns, a certain text is interpreted as a more or a less weighty claim, as a legally binding rule, as a normatively binding standard, as a technical standard or as just a convenient habit.

We should not think that there is one correct pattern. As I argued in my earlier article,Footnote ⁴⁹ opinio juris should be understood in its plural form. There are different opiniones juris: official claims that propose different ordering patterns and which function as maps. The main function of such maps is not merely to represent reality. Maps typically highlight some elements at the expense of others. The cyclist map ignores the highways; underground maps emphasise the underground stations and may even be extremely unrealistic in the sense that they distort the physical distance between stations. The value of such maps does not depend on their realism but on the use – the ‘point’ – of such a map. The investigation of state practice, but also of CIL as a whole, can best be seen as a map drawn on a transparent sheet that is laid over the abundance of official documents and texts, by means of which this legal material is ordered in different ways.

All this implies that the claim cannot be justified that there is an objective examination of state practice and a neutral ‘identification’ of rules that precedes their interpretation. The transparent sheet cannot be removed without risking being plunged into a bewildering multitude of incomprehensible texts. But, and this is important, this does not imply that judgments are completely arbitrary. The courts are not ‘objective’, but neither are they ‘subjective’. They do not deal exclusively with the factual, but nor are they condemned to the realm of normative ideals. Their judgments are never completely apologetic or utopian, because the interpretation of legal texts is not exclusively limited to what is, nor to what ought to be. The task of courts is to choose a path among possible worlds.

This implies a somewhat nuanced answer to the common objection against CIL as mere fantasy. It seems exaggerated to speak of judicial discretion in the strong sense of the word, according to which the judge is simply not bound by any rules. In fact, even in the advisory opinions where the ICJ is just required to answer very open questions, there is always plenty of legal material that may or may not support the opinion. Courts do not fantasise. But neither are they unreservedly bound by that material. The texts are not determinate strongholds. Their weight is (co-)determined by the patterned transparent sheets. The plurality of these sheets leaves room for choice.Footnote ⁵⁰

1 Introduction

Most social groups have rulesFootnote ¹ about how the people in that group should behave. In many cases, these rules develop from custom; they are called ‘customary rules’. New customary rules arise from two elements: (i) the repeated practice of certain acts by the people in a group (diuturnitas); and (ii) a shared expectation that this practice will continue. This expectation leads people to sense or feel that they should act in a certain way. That is, people continue to engage in a particular practice because they believe they are obligated to do so (opinio juris sive necessitatis, or simply opinio juris). By contrast, practice without a sense of obligation leads to habits rather than customary rules.Footnote ²

Customary rules have become less important in domestic law, but the opposite is true in international law. International lawyers frequently rely on customary international law (CIL), and many of the most important rules of international law have come from CIL. The prohibition of genocide, torture, slavery, and piracy are relevant examples.Footnote ³ We can also see the importance of CIL for topics not yet regulated by treaties and topics addressed by treaties that are not universally ratified. The Law of the Sea ConventionFootnote ⁴ is a good example of this. The Convention has helped codify CIL rules already in place and blended them with treaty rules, smoothing some of the differences between CIL and non-CIL rules.Footnote ⁵

As a source of international law, CIL is not only important but also complex. Like international law in general, CIL results from a bottom-up process, where subjects of the law are also its makers. To complicate matters further, CIL develops informally and by implication. In CIL, unlike treaty law, there is no standard way for states to gather and decide on creating new rules. Instead, CIL rules seem to appear out of nowhere; they are just here, and it is sometimes unclear how long they have been with us.Footnote ⁶

Another complication that affects CIL is the well-known chronological paradox. Despite its name, the paradox is not merely chronological: it raises a broad conceptual issue that threatens the coherence of (customary) international law. The paradox lies in the fact that for a new CIL rule to be created states must believe that the law already requires them to act in a certain way (opinio juris), but until the rule is created they are not legally required to do so as the rule establishing such a requirement has yet to exist.Footnote ⁷

Since there does not seem to be any way out of the chronological paradox’s vicious circle, the making of a new CIL rule would seem impossible, or, at the very least, our understanding of CIL would appear incoherent. Given that we know CIL rules do exist, their creation is not impossible, so we must entertain the possibility that our understanding of CIL is incoherent.Footnote ⁸ This is critical, as the legitimacy of a legal system depends in part on its coherence. If legitimacy is what makes us want to carry out our legal obligations, then coherence – the intelligibility of such obligations – is a requirement for legitimacy. After all, we can only effectively comply with what we understand.

Given that CIL rules are being used more and more against states, the chronological paradox is both a theoretical and a practical problem.Footnote ⁹ If the paradox is true (and, thus, our understanding of CIL is incoherent), states may be less inclined to follow CIL rules, which would be a blow to international law. Given how critical the paradox is, there is reason to think that the International Law Commission (ILC) would have directly addressed it in its 2018 report on identifying CIL.Footnote ¹⁰ Unfortunately, this was not the case. Ignoring the paradox’s significance is problematic, as it casts a large shadow over CIL. In a landscape where CIL is a primary source of international law, practitioners should not be left stumbling in the dark.Footnote ¹¹

This chapter tries to shed some light on this topic by addressing the paradox in a way that ensures the coherence, and thus legitimacy, of (customary) international law. It does so by offering an interpretation of CIL that accounts for the creation of CIL rules through distinct, non-paradoxical steps. The chapter argues that the belief in a legal obligation (opinio juris) derives from the general principle of good faith. Good faith leads to legal obligations, which compel a subgroup of states to engage in specific behaviour. Then, as a result of this subgroup’s repeated behaviour, a new CIL rule develops, obligating the entire community of states to follow that behaviour. In elucidating the shift from good faith to legal obligations and from legal obligations to CIL rules, this chapter draws on interpretivism, social ontology, and contemporary research on constitutive rules.

In terms of structure, the chapter is divided into four sections. Following this introductory section, Section 2 discusses CIL and the chronological paradox. Section 3 examines constructivism in legal interpretation. Section 4 then focuses on the application of rules and on constitutive rules, explaining the role of good faith and how constitutive (meta)rules create new CIL rules. Section 5 concludes the chapter by reaffirming earlier conclusions.

2 CIL and Its Paradox

Article 38(1)(b) of the Statute of the International Court of Justice (ICJ)Footnote ¹² states that CIL rules evince general practices that are accepted as law. The Statute thus refers to the two constituent elements of customary rules mentioned in Section 1 – namely, (i) the existence of a general practice in the form of repeated acts (diuturnitas), and (ii) performance of these acts in the belief that they are required by law (opinio juris).

Scholarship traditionally considers practice (diuturnitas) to refer to the practice of states. There have been fruitful discussions about the contribution of non-state actors to CIL.Footnote ¹³ For instance, the ILC has recognised that ‘[i]n certain cases, the practice of international organizations also contributes to the formation, or expression, of rules of customary international law’.Footnote ¹⁴ For the purposes of this chapter, the question of whether non-state actors’ behaviour qualifies as ‘practice’ is beside the point; although an interesting debate, it has no bearing on the chapter’s explanation of how CIL rules are created. Thus, although the chapter addresses state practice, it can also be read as encompassing non-state practice.

Acceptance as law (opinio juris) means that states believe a specific action is required under the law. In other words, states believe that a particular practice is legally obligatory. It should be pointed out that the term ‘obligation’ is used in this chapter to designate a basic normative or deontic concept.Footnote ¹⁵ When we say that a particular behaviour is obligatory or required, or that someone must act or is responsible for acting, we mean that there is an obligation to act in a certain way. Some obligations, such as the obligation to pay taxes, are imposed by the law and legal rules. These are legal obligations. However, there are also non-legal obligations, like the obligation to honour a promise made to a friend. It is worth noting that obligations can be used as a basis from which to derive prohibitions or permissions.Footnote ¹⁶ A prohibition against using force, for example, is an obligation not to use force. Being permitted to use force in self-defence is tantamount to not being prohibited from using force in self-defence (a non-obligation not to use force in self-defence). Thus, while continuing to refer only to obligations, we use the term broadly to cover permissions and prohibitions, too.

When it comes to CIL, states act in the belief they are bound by a legal obligation (opinio juris). By belief, we to refer to a general acceptance of something as being the case.Footnote ¹⁷ Different beliefs move us in different ways. Some beliefs are about things we know for sure, while others are about things we only vaguely sense or feel. No matter how strong or weak, these are all beliefs. Besides, believing in something does not mean we have to think about it or do it all the time. For example, most of us believe that the moon is not made of cheese, but we do not usually think about it. Let it be said that there is nothing controversial about attributing beliefs to states: following Cassese, we can say that it is humans acting on behalf of states who believe that certain acts are obligatory.Footnote ¹⁸

Opinio juris allows us to distinguish practices that are merely habitual from those that create normative expectations in the community and consequently lead to the creation of CIL rules.Footnote ¹⁹ It should be borne in mind that what we are referring to here is the belief that a practice is required by law. States feel compelled to perform these acts because they consider them to be legally obligatory, not for any other, non-legal reason, no matter how strong it may be. As stated in Section 1, the chronological paradox suggests that CIL is caught in a vicious circle: to create a new CIL rule, states must believe that the law obligates them to act in a certain way (opinio juris), yet until this CIL rule is created the act in question is not required by law.Footnote ²⁰

To add colour to this explanation, let us imagine that a group of wealthy states begins donating vaccines to developing states during a pandemic. Suppose this practice continues to the point where states believe that they are under a legal obligation to donate vaccines. Lawyers then identify a new CIL rule, according to which wealthy states are obligated to donate vaccines to developing states during pandemics. The problem is that if this new ‘vaccine donation CIL rule’ exists simply because states believe the practice to be legally obligatory, then the rule cannot as such impose the obligation to donate vaccines. For it actually to be a legal obligation, a legal rule imposing the obligation is needed. As Bradley puts it: ‘If State practices do not become binding as CIL until the States involved act out of a sense of legal obligation, how do the States develop that sense of legal obligation in the first place?’Footnote ²¹ In our example, if wealthy states are legally obligated to donate vaccines only by virtue of a CIL rule, how did they come to believe that the law obligates them to donate vaccines in the first place?

The paradox is disconcerting, because if it correctly accounts for the creation of new CIL rules, then CIL is fated to chase its tail. Such circular reasoning is problematic: it uses its conclusion as a premise, while giving us no reason to accept its premise other than that we already believe its conclusion. We have to assume that the law requires specific behaviour on the part of states (there must be a legal obligation) before we can conclude that a given CIL rule exists. But to assume that the law requires such behaviour of states (i.e. that there is indeed a legal obligation), we need a pre-existing CIL rule commanding such behaviour (i.e. we need a CIL rule to impose that legal obligation).

Gény gives a straightforward (but unsatisfactory) answer to the paradox. He says that states are wrong to think they are legally required to act: their opinio juris is mistaken.Footnote ²² That is, states have – or feign – a false view about what is happening,Footnote ²³ whereas there is no obligation at all. Gény says that when this mistake is made repeatedly by a lot of states, it generates mutual expectations that turn into a legal obligation. Gény’s response admittedly short-circuits the paradox, but in avoiding one problem it falls into another, for if states are mistaken, then CIL is incoherent and its entire structure based on a delusion. As exposed in Section 1, such a workaround falls short of what this chapter sets out to do, which is to address the paradox in a way that preserves the coherence, and thus the legitimacy, of (customary) international law.

Some authors have made intriguing attempts to re-imagine opinio juris as more than just a description of the law, arguing that it helps to create the law. To put it differently, these authors claim that opinio juris goes beyond a descriptive stance on the existence of legal obligations, because it also plays a prescriptive role in determining the content of these legal obligations. In Westerman’s words: ‘Opinio juris can better be regarded as a map, which represents the law by selecting and highlighting those aspects that are deemed important. It describes and prescribes at the same time.’Footnote ²⁴ When states believe that a certain act is legally obligatory, they are not simply describing a pre-existing reality. Their belief is effectively building reality as we know it. This – the constructive role of our collective beliefs – is a topic central to social ontology; as a field of study, it investigates how the combined action of our minds can shape the world around us. According to social ontology, our minds are not (just) describing reality; they are projecting meaning onto it.Footnote ²⁵ For example, when people believe that a river marks the frontier between two communities, they are not only acknowledging that fact but also creating it. If it were not for people collectively believing that the river is a frontier, it would be just another watercourse.

The depiction of opinio juris as a belief with both descriptive and prescriptive functions does not contradict traditional views on CIL. In fact, the ILC’s explanation of acceptance as law (opinio juris) corresponds to social ontology’s account of collective acceptance. The ILC explains that acceptance as law implies there is a ‘subjective’ or ‘psychological’ dimension to the ‘binding character of the practice in question’, in the sense that ‘the practice must be undertaken with a sense of legal right or obligation.’Footnote ²⁶ According to social ontology, collective acceptance is when the members of a social group come to share and maintain a relevant social attitude (a ‘we-attitude’) towards a certain practice.Footnote ²⁷ Like the constructive role of beliefs in the river-frontier example mentioned above, when states believe they have an obligation to engage in a particular practice, this belief is both descriptive and prescriptive. It should be added that participants must be committed to what they believe, though their belief can range from enthusiastic endorsement to simply going along with the group.Footnote ²⁸ As a result, the lines between passive acceptance and active support or desire are frequently blurred.

3 Constructive Interpretation

In Section 2, we discussed the two constituent elements of CIL, dissected the chronological paradox, and began to discuss how our collective beliefs can shape the social world. We saw that the crux of the paradox lies in the manner in which a new CIL rule is created: how can the existence of a CIL rule be dependent on a condition (belief in a legal obligation – opinio juris) that appears to be itself dependent on an already existing CIL rule? The challenging feature of the chronological paradox is thus the transition from the first point in time when a CIL rule does not yet exist to a second point in time when it does. To understand the transition between these two points, we must first understand how CIL rules are identified.

The identification of a CIL rule involves gathering evidence that its two constituent elements – repeated practice (diuturnitas) and a belief in legal obligation (opinio juris) – are present. Such evidence is sought in prior pronouncements and empirical data relating to state behaviour. Through observation of relevant materials (wherever these materials are found), the aim is to determine whether a specific CIL rule exists. If sufficient evidence of the two constituent elements is found, a new CIL rule is said to exist. If the evidence is insufficient, there is (as yet) no new CIL rule.

The search for corroborative materials should not be thought of as ‘gold-digging’ for evidence.Footnote ²⁹ Far from being based on neutral observation, the identification of CIL rules is a highly interpretive activity. As philosophers have long recognised, no evidence is ‘given’ neutrally, because all observation is theory-laden:Footnote ³⁰ the theoretical framework of the observer/interpreter determines the meaning of what they observe. Likewise, how people choose, see, and judge evidence is affected by the theories they hold.Footnote ³¹ In a legal context, as explained by Hart, ‘situations do not await us neatly labelled, creased, and folded; nor is their legal classification written on them to be simply read off by the judge.’Footnote ³² On the contrary, evidentiary materials call for purposeful interpretation: it is the responsibility of the interpreter to determine the import of the words and actions.Footnote ³³

The question therefore arises as to what role and purpose interpretation has in relation to CIL rules. There are at least two possible answers to this question, for legal interpretation may be of two kinds: descriptive or constructive. The former simply describes what the law is and what rules exist. The latter not only tells us what the law is but also creates the law and its rules. The descriptivist viewpoint has ‘realist’Footnote ³⁴ undertones, because it presupposes the existence of a legal reality independent of the interpreter, whose task is therefore merely to point out answers that are already there. By contrast, the constructivist viewpoint is based on the premise that legal interpretation is more than just a search for answers; it considers the interpreter to be engaging in an act of creation when assigning meaning to what they interpret.Footnote ³⁵

This chapter discounts the descriptivist point of view for a number of reasons,Footnote ³⁶ one of which is that descriptivism ignores the fact that observation is theory-laden. As previously stated, all observation, including interpretation, is coloured by the observer’s theoretical bias. Consequently, observation (or interpretation) can never be purely descriptive; it always has, to some extent, a constitutive role. Furthermore, the descriptivist viewpoint is unable to account for the necessary role argumentation plays in the interpreter’s efforts to solve ‘hard cases’.Footnote ³⁷ Consider the discussions currently taking place over whether there is a customary rule that allows states to intervene in the territories of other states to stop human rights violations;Footnote ³⁸ whether there is a level of gravity below which small-scale armed attacks are not considered acts of aggression under international law;Footnote ³⁹ and whether it is illegal for states to conduct military exercises or manoeuvres in foreign exclusive economic zones.Footnote ⁴⁰ The answers to these questions are not a ‘given’, but rather depend on the quality of the interpreters’ arguments. In other words, the ideas of those who interpret the law shape the contents of international law and international legal rules.

Dworkin took a similar constructivist approach to legal interpretation in saying that interpretation is a creative process in which meaning is imposed on the object of interpretation to make it the best instantiation of a given form or genre.Footnote ⁴¹ However, he did not think interpreters should be able to interpret sources however they wished. Although interpretation must allow for innovation to ensure that the object as interpreted best serves its purpose, its scope is limited by the object’s history.Footnote ⁴² Thus, legal interpretation is Janus-faced, looking both backwards and forwards.Footnote ⁴³ While based on the observation of pre-existing materials, it is also creative in that it is not a mere reproduction of those materials; rather, it creates something new by giving them meaning. This is equally true of CIL interpretation, for once evidence of the two constituent elements of CIL has been established, CIL interpretation fulfils a constructive role by shaping and determining its content.

It is important to note that we can still talk about ‘interpreting’ CIL even when the evidence of its two constituent elements is not in textual form. True, interpretive activity in the legal sphere is often directed at pieces of text, as when we extract rules from the text of a treaty. This, for instance, happens when we read Article 8 of the Rome StatuteFootnote ⁴⁴ to determine the content of the rule punishing aggression. But interpretation goes beyond textual hermeneutics. It is possible to distinguish between two types of interpretation – (i) source interpretation, which is the interpretation of (formal and material) sources of law; and (ii) rule-to-case interpretation, which is the classification of a case according to the conditions under which a given rule is applicable to it.Footnote ⁴⁵ In both cases, interpretation may be directed at unwritten materials. For example, when lawyers identify a new CIL rule on the basis of empirical evidence (source interpretation), or determine, on the basis of visual evidence of armed conflict, that the crime of aggression has been committed (rule-to-case interpretation), they are relying on unwritten materials.

Both the ICJ and the ILC believe that CIL is susceptible to interpretation. In Nicaragua, the ICJ held that rules derived from CIL could be distinguished from those derived from international treaties by ‘reference to the methods of interpretation and application.’Footnote ⁴⁶ Likewise, in North Sea Continental Shelf Judge Tanaka asserted that ‘[c]ustomary law … requires precision and completion about its content. This task, in its nature being interpretative, would be incumbent upon the Court.’Footnote ⁴⁷ The ILC has taken a similar line: when commenting on materials that may be consulted to help identify CIL, it stated that specific ‘texts may assist in collecting, synthesizing or interpreting practice relevant to the identification of customary international law’.Footnote ⁴⁸ It furthermore explained that ‘[e]ach of the forms [of state practice] listed is to be interpreted broadly to reflect the multiple and diverse ways in which states act and react’.Footnote ⁴⁹ Therefore, it cannot be considered controversial to claim that CIL is interpretable.Footnote ⁵⁰

To conclude this discussion of the interpretive nature of identifying CIL rules, it is crucial to keep in mind that there is no absolute separation in how evidence for the two CIL elements is interpreted. Although the ICJ held in the Asylum case that diuturnitas and opinio juris should be determined separately,Footnote ⁵¹ this view has gradually been replaced by one that allows evidence of both constituent elements to be extracted from the same act. D’Aspremont discerned signs of this change in the Gulf of Maine and Nicaragua cases.Footnote ⁵² The most recent rebuttal of the ICJ’s initial position came from the ILC in its 2018 Draft Conclusions. When defining what materials may serve as evidence of repeated practice and the belief that such practice is legally required, the ILC acknowledged that one and the same act may provide evidence of both.Footnote ⁵³

4 The Formation of CIL Rules

This section will break down the formation of CIL rules into distinct steps. It is a process that begins with the principle of good faith, which leads to legal obligations, and these legal obligations in turn contribute to the development of a new CIL rule. But before we can discuss these steps, some general ideas must first be addressed – namely, rule applicability and application, constitutive rules, and rules about rules (metarules). Although these matters might not seem directly related to the chronological paradox, their elucidation is important to further discussion of the paradox. Given that problems are best solved when broken down into their most basic parts and then put it back together again, this section will focus on deconstructing and rebuilding the problem.

5 Final Remarks

This chapter argued that CIL rules result from applying a constitutive metarule. A constitutive metarule is a metarule (a rule about rules) that creates another rule when applied. When the constitutive CIL metarule (which we call M) applies, it creates an individual CIL rule (referred to as R). This chapter has proposed two alternative explanations for how a belief in a legal obligation (opinio juris) can arise prior to the imposition of such an obligation by a CIL rule. In doing so, it assumes in both cases that states are correct in believing they are legally obligated. So there is not just a belief in a legal obligation; there actually is a legal obligation.

The first explanation presented the obligation as a basic social entity based directly on what states believe. Some may be satisfied with this explanation. However, others might criticise it for failing to bridge the gap between ‘is’ and ‘ought’. It could also be argued that this first explanation does not clarify how this basic social entity can impose a legal obligation. For these reasons, this chapter laid aside this first explanation and went on to develop an alternative explanation.

The second explanation addresses the chronological paradox by breaking down the creation of a new CIL rule R into three non-contradictory steps:

• Step 1: A subgroup of states starts to act in a certain way. Their behaviour leads to legitimate expectations that they will continue acting that way. This subgroup of states is now obligated to continue acting pursuant to a rule derived from the principle of good faith (we call this rule L). However, because this obligation arises from a rule (L) outside general international law, its enforceability is questionable.

• Step 2: The practice of this subgroup of states spreads among other states to the point where it becomes a general practice accepted as law. So, the two CIL elements – diuturnitas and opinio juris – are present. Therefore, the CIL metarule, M, is now applicable.

• Step 3: M applies, thus creating a specific CIL rule, R. This CIL rule R imposes an obligation to act on all states except for persistent objectors (unless R is a jus cogens rule, in which case it applies even to persistent objectors).

The creation of a new CIL rule as a three-step process allows us to unravel the apparent circularity of the chronological paradox. It also allows the creation of CIL rules to be explained in such a way as not to jeopardise the coherence (and legitimacy) of (customary) international law.

Some of the claims made in this chapter might seem controversial, because it is unusual to think of CIL rules as being created through the application of a (meta)rule. We are more inclined to think that it is a matter of finding already formed CIL rules rather than creating them. But that intuition is not incompatible with the contention made in this chapter. For when we find CIL rules, we are doing more than just describing; we are giving meaning to the things being interpreted. In the case of CIL, these things are the proof of practice (diuturnitas) and belief in a legal obligation (opinio juris). We should consider the CIL metarule as an unspoken rule that operates behind the scenes. Even if its action is unannounced, this metarule helps international lawyers interpret the law to identify individual CIL rules. The point is that for lawyers in their daily work the three steps set out above occur simultaneously, giving the impression that CIL rules are paradoxical. By looking at CIL analytically, however, we have been able to pinpoint these separate steps and show that CIL is not flawed by circular reasoning.

Like the rules of chess, the CIL metarule operates in the background. When experts play chess, they do not have to keep referring to the rules and reminding themselves that, for example they are using the checkmate rule: when they look at the chessboard, they know it is checkmate. The same is true in international law, with the important difference, however, that law is an argumentative activity.Footnote ⁹⁸ Unlike chess, it is not an activity where all the rules are laid down categorically. On the contrary, the law involves deploying and disputing the very rules that constitute it. Part of what it means to be a lawyer is the ability to understand and question the law itself. In this respect, one of the goals of legal philosophy is to make the implicit background rules of legal activity explicit.

1 Introduction

When faced with the inevitable task of interpreting customary international law (CIL), what should a court do, and what should it consider? Should the court engage in an ‘inductive’ process of sifting through available evidence of state practice and opinio juris, or a deductive process designed to reason logically from principles embedded in the rule? Should the court invoke something like the rules of treaty interpretation with their focus on good faith, ordinary meaning, context, and object and purpose? How should it weigh available evidence of state practice, opinio juris, and prior interpretations – both its own and that of others? How concerned should the court be about fitting its interpretation into the broader corpus of international law?

This is where international law doctrine – particularly concerning sources – falls short. Figuring out how to interpret and apply custom requires a theory of custom. What is customary international law? Why is it worthy of being followed? How does custom acquire normative and legal authority? These are difficult and contentious questions we (and certainly courts) usually try to avoid.Footnote ¹ The doctrine of sources does not answer them. Article 38 of the Statute of the International Court of Justice (arguably) lists the formal sources of international law – treaties, customary international law, and general principles. Its formulation ‘international custom, as evidence of a general practice accepted as law’Footnote ² has been interpreted as implying a two-element test, requiring evidence of both state practice and opinio juris.Footnote ³ But it provides no account of why those elements would, together, create law or be deserving of normative authority. The International Law Commission (ILC) also studiously avoided these questions in its Draft Conclusions on Identification of Customary International Law.Footnote ⁴

This is perhaps not surprising. These theoretical questions are difficult and contentious. International actors disagree over the answers, sometimes deeply. And Article 38 and the two-element test might best be seen as an attempt to bridge such theoretical disputes. It encodes formal sources that we agree count as international law, without encoding why. Focused on method rather than theory, the doctrine provides a discursive convention – an agreed-upon framework for structuring claims about rules that is capacious enough to accommodate a range of views on where they come from. It acts as a normative compromise that allows us to operate on the basis of overlapping consensus while leaving room for a certain amount of normative pluralism. If Article 38 and the two-element test count as a rulebook,Footnote ⁵ it is a rulebook on how to argue.

Questions about interpretation, however, reveal the instability of this compromise. We simply cannot answer questions about how rules should be interpreted without some theory as to why certain interpretations would or would not be legitimate. And for an interpretation to be legitimate it must be able to trace its legitimacy to that of the international law source being interpreted and/or the legitimacy of the particular interpreter. Figuring out both requires looking behind the list of sources in Article 38 and asking why we treat each one as a source of normative authority. It requires us to focus on the normative stories we tell.

Such an inquiry, though, reveals that there is not just one story explaining why custom should be a source of law, but multiple stories, each legitimated in very different ways. In fact, if we started not with Article 38 but with our accounts of international law’s legitimacy, we would find, as I argue below, that what we call ‘custom’ may represent or draw from at least three different sources of law,Footnote ⁶ which I will refer to as ‘Negotiated Law’, ‘Legislated Law’, and ‘Adjudicated Law’.Footnote ⁷ Although we refer to nontreaty rules derived from each as ‘custom’,Footnote ⁸ they draw on different sources of legitimacy, operate according to different logics, dictate different methods of interpretation, and favour different methods for resolving disputes. Answering questions about any of those aspects of the rule thus requires us first to figure out the sort of custom we are talking about. The types of justification we start with dictate the answer to a range of other questions we might have about the rules emanating from them. And those debating how custom should be interpreted, I argue, are often arguing about different sources of law. Their disagreement is not over how custom should be interpreted, but over the ultimate authority of particular international law rules.

Some readers may bristle at the suggestion that these are distinct ‘sources’ of law. Do we really need to upend our categories of treaty, custom, and general principles? To the extent, though, that the rules seem to be emanating from entirely distinct sources of legitimacy and operating according to contradictory logics, it is hard to describe them as the same in any way other than form. As already noted, formal lip service to the features in Article 38 may play a role in anchoring discussions, but it is unlikely to capture deeper agreement about where these rules come from or what they represent.Footnote ⁹ These three models are meant to capture the true sources of these rules’ authority.Footnote ¹⁰ That said, even if one is unwilling to embrace the idea of plural sources of custom, recognizing that there may be different types of custom may help clarify the stakes involved in debates over interpretation and in interpretive choices.

This chapter starts in Section 2 with a reconstruction of international law’s sources built not on the forms described in Article 38 but instead on the legitimating stories we tell. Section 2 lays out three such stories and the types of law they describe. These three types of law – Negotiated Law, Legislated Law, and Adjudicated Law – each draw on distinct notions of legitimacy, follow distinct logics, and instantiate distinct values. Section 2 describes the very different notions of ‘custom’ emanating from each of these sources. Section 3 looks at each of these three new sources from the standpoint of interpretation. How are each of these sources interpreted? Section 3 explains how the justificatory story told about each of these sources flows through interpretation, requiring different approaches attentive to different concerns. Again, interpreting ‘custom’ will require very different approaches and operations depending on its ultimate source and justificatory logic. Finally, Section 4 shifts the vantage point to courts. How should courts approach the task of interpreting each of these sources? Each type of custom, this section argues, dictates not only its own interpretive rules but the specific role of the court in applying them and the weight to be given to the court’s judgments in the future. The hope is that this reconstruction of international law’s sources, including the identification of different types of custom, can put old arguments about custom and its interpretation in a new light, explain the intractability of certain fights, anchor phenomena that others have observed, and clarify the real stakes behind differing methods of interpretation.

2 Three Models of Custom

Article 38 largely describes international law’s forms – written agreements, customary practices, accepted principles. What if we defined international law’s sources instead according to their sources of legitimacy? What if we started by asking why we give international law rules normative force at all, listening more carefully to the explicit and implicit legitimating stories we tell? This section lays out three prominent stories we tell that cut across form, but for the purposes of this volume explains why nontreaty rulesFootnote ¹¹ might carry the force of law. Each of the stories produces a different model of law – Negotiated Law, Legislated Law, and Adjudicated Law – drawing legitimacy from different sources, instantiating different values, and following its own logic.

3 Interpretive Paths

Negotiated Law, Legislated Law, and Adjudicated Law each have their own sources of legitimacy and operate according to attendant logics. Each reflects a coherent story as to why a rule developed and operating in a particular way should carry normative authority, why it deserves to be treated as law. For interpretations of those rules to benefit from those rules’ legitimacy, they must be able to fit within those same stories. What legitimates the rule should also legitimate interpretation. This means that interpretation will look different, involving different processes and considerations, depending on the asserted source of the rule. These divergent methods of interpretation also help explain why the interpretability of custom remains so controversial: those taking opposing views may simply be focused on different sources, all unfortunately labelled ‘custom’.

4 The Quandary of Courts

Negotiated Law, Legislated Law, and Adjudicated Law are thus best seen as distinct sources, with their own legitimating logics, favoring different interpretive techniques. These differences highlight the problem for courts of interpreting custom (or really any source of international law). Although ostensibly focused on one formal source in Article 38 – custom – courts are faced with a choice between divergent paths, a fork in the road, that will dictate the factors they should consider and the weight they should give them. These choices can be stark, resulting in contradictory interpretations of what custom requires. As such, courts need a theory, whether explicit or implicit, of what ‘custom’ is and where it comes from before they can interpret any particular rule before them.

The challenge for courts can be illustrated well by the problem of coherence. We might want courts to worry about coherence,Footnote ⁷⁹ but coherence with what? The different possible sources described here – Negotiated Law, Legislated Law, and Adjudicated Law – point to different concepts of coherence favoring different outcomes in specific cases. Here, we can go in the opposite order, moving from the source easiest for courts to interpret to the hardest.

Courts will have the easiest time with Adjudicated Law. Adjudicated Law maps what courts do (or at least imagine themselves doing) well: asking judges (or other decision-makers) to resolve arguments about the meaning and requirements of the law by reasoning their way to definitive answers. It leans on their (perceived) instincts, talents, and expertise.

For the court engaged with Adjudicated Law, the demands of coherence will point first and foremost to what might be called ‘courtroom coherence’. Courts working within an Adjudicated Law model need to worry that like cases be treated alike and that litigants be able to predict the rules that will apply to them. These are core factors legitimating a court’s authority to explicate and develop legal rules, which only become heightened in international criminal cases where they may merge with human rights and due process. Courtroom coherence grows out of the model’s emphasis on legitimacy of reasoning and expertise. Both of those would be undercut if rules were interpreted differently in each case: if the court’s legal reasoning is so solid, why can it not convince itself? If the court is so expert in the law, why does it have so much trouble finding the right rule? But courtroom coherence also draws from Adjudicated Law’s emphasis on justice and binary choices of right and wrong, winner and loser. It becomes much harder to argue that one answer is the just one in any given case if that answer’s justice seems so unclear over time.Footnote ⁸⁰

In sum, Adjudicated Law’s emphasis on courtroom coherence and the expectations of litigants before them should lead courts to favour interpretations concordant with their own prior rulings, to follow a form of precedent, even at the expense of developments elsewhere in international law or renegotiation or clarification by states. When investment tribunals do just this, rightly or wrongly, Adjudicated Law seems to be the model they have in mind.

Legislated Law, drawing on a story of ratification, common understanding, and the demands of cooperation and planning, also favour the coherence that comes from predictable rules. Focused, though, on the quasi-legislative nature of international agreement, Legislated Law will favour transsubstantive, transforum predictability – the coherence of systemic integration. Legislated Law is meant to avoid the uncertainty of Negotiated Law and the ex-post justice of Adjudicated Law by creating clear frameworks for action that can be interpreted in common ways with as little friction between them as possible. Legislated Law values coordination over conflict, settled rules over settlement. Fragmentation undercuts these goals and may even seem illegitimate. A court placing its own precedent ahead of the will of the international community becomes a judicial despot. Legislated Law will thus favour common rules for all, accepted by all, which can integrate rules across courts and regimes into a single doctrine of international law.

In a vacuum then, courts applying Legislated Law would still be expected to hew to prior rulings. But that expectation would be based on the assumption that the court was correct in its prior interpretation, not on notions of litigant fairness. A single court’s interpretation should remain revisable in the face of evidence of other courts having arrived at a different interpretation or in light of developments elsewhere in international law. Legislated Law argues in favour of transjudicial dialogue and drawing from one set of rules to another, whether environmental law into trade or human rights into humanitarian law.

Negotiated Law presents the most difficult problems for courts. As noted in Sections 2.3 and 2.4, the logics of Negotiated Law and of courts are in some opposition.Footnote ⁸¹ Courts are asked to decide in favour of one party or another, to decide that one actor was right and the other was wrong; Negotiated Law prefers negotiated settlements that leave room for future negotiation and wrangling. Courts are asked to provide justice; Negotiated Law is looked to provide a modicum of peace. Courts lean on delegated responsibility to the law; Negotiated Law leans on the autonomy of parties to order their affairs. To some extent, courts are faced with a problem like Heisenberg’s: deciding a case based on Negotiated Law risks transforming it, giving its rules a level of certainty that wrangling within that model would not have produced.

If courts are to be attentive to Negotiated Law, if they believe that Negotiated Law is the true source of the rules before them, then they must be very careful not to overstep their role. Courts should be minimalist in their decision-making and treat their decisions as ad hoc and revisable. They should revisit state practice in each new case regarding the rule at issue to see whether their prior interpretation has been vindicated or rejected.Footnote ⁸² Here, the coherence sought is negotiating-table coherence. The goal for courts interpreting Negotiated Law is to maintain states’ expectations about the shape of the negotiating space, about what has been negotiated and what has not. A court should not be able, like an interloper, to stick its nose in ongoing negotiations (which in this model are all the time and endemic) and throw things in favour of one side or another.

An analog to these problematic, divergent choices can be seen in the WTO Dispute Settlement Understanding (DSU). While largely a matter of treaty interpretation, the problem is the same. DSU 3.2 provides that:

In other words, WTO panels and the Appellate Body should stick to negotiated agreements, and only those agreements, but provide security and predictability (a justification for precedent), all while taking account of customary rules of interpretation of public international law, which of course might require looking at other, external agreements and general international law.Footnote ⁸⁴ The impossibility and incompatibility of doing all of those at the same time is evident in the Appellate Body’s attempts to interpret rules regarding antidumping duties and zeroing, where the Appellate Body, seeking courtroom predictability, adopted a doctrine of precedent, policed by a cogent reasons standard.Footnote ⁸⁵ Simple, the Appellate Body thought, we will apply the same zeroing rule regardless of plaintiff and regardless of defendant. But to the United States, this position regarding both the rule and the Appellate Body’s authority was simply wrong. In response, the United States declared war on the body, drawing on a Negotiated Law model to argue that zeroing was a gap in the agreements purposely left for future negotiation and explicitly questioning the Appellate Body’s authority to fill it.Footnote ⁸⁶

Fair and equitable treatment in investment law provides another example. Faced with a question about the scope and meaning of the concept,Footnote ⁸⁷ an arbitral panel is left with a series of choices:Footnote ⁸⁸ if it wants to be true to the Negotiated Law model, it will adopt a minimalist view of role and precedent, see its interpretations as mere proffers, and be extraordinarily attentive to state responses to these decisions, including attempts to renegotiate bilateral investment treaties. This, though, would come at the cost of doctrinal coherence or courtroom coherence. If the panel instead worries most about its authority as court-like tribunal and about courtroom coherence, it will strive to treat like cases alike and stick to its own precedent, potentially at the cost of both systemic integration and negotiating space. If, though, the panel is concerned about systemic or doctrinal coherence, it might eschew the consistency of its own opinions in favour of looking to jurisprudence constante or acquis, or even beyond to foster systemic integration by looking to rules about the environment, human rights, and sovereignty. In this case, the panel could rely neither on Negotiated Law bases of legitimacy nor on its delegated authority. Instead, it would have to rely on the authority of the international community. Systemic integration might be achieved, but only at the cost of treating like cases alike, respecting specific negotiated deals, and respecting space for future deals (see Table 2).

Coherence, ironically, thus represents a fragmented choice. Different courts will take different views on these issues, but these decisions are always contested and contestable. What is often only recognized by critics, though, is that these choices are foundationally not about coherence or methods of interpretation, but about justifications of authority. Even if only implicit, the legitimating source of a rule must be identified before the proper interpretative tools can be deployed.

5 Conclusion

Debates over the proper interpretation of customary international law reveal the limits of doctrinal consensus. Methods of interpretation draw on the authority of the particular interpreter and the authority of the rules being interpreted. As such, the choice of methods must be sensitive to our accounts of each. But surfacing those accounts by teasing out the stories we tell to justify customary international law rules reveals the deep theoretical disagreements buried beneath our doctrine of sources. Multiple, competing, perhaps even contradictory concepts of customary international law are in common use; what we call ‘custom’ might emanate from any of those sources of authority. It is no wonder that we disagree on how to interpret it. Interpretive debates reflect the varied sources we have agreed to group under the label of ‘custom’ and the deep normative pluralism undergirding the agreed formal sources of international law. Competing justificatory accounts of customary international law that formal doctrine seeks to ignore resurface in these debates. Interpretation becomes the battleground on which our justificatory accounts of custom are forced into combat.

1 Uncertainty and Change: The Dynamic Character of CIL

The uncertain character of CIL has been discussed time and time again among academics and practitioners. To most of them, the uncertain character of CIL entails a twofold defect: first, uncertainty is perceived with respect to the identification of the rule, since we may not know whether there is a valid legal rule,Footnote ¹ and second, uncertainty is perceived with respect to the content of the rule, since we may not know the precise meaning of the rule.Footnote ²

Thus, uncertainty is viewed as the basic reason for the lack of coherence and consistency attributed to CIL rules. In turn, it is argued that uncertainty is a doctrinal issue discussed among scholars but not among judges. However, judges will inevitably deal with uncertainty, even indirectly, when they have to assess the meaning of the rule to apply in the case at hand. Uncertainty will in this respect lead to unpredictability (and, consequently, even greater uncertainty) when different legal meanings taken from a range of probabilities are assigned to a CIL rule. But if uncertainty is taken to mean ambiguity – i.e. different interpretive approaches to an applicable rule – then, by induction, there can be no evolution of CIL, which leads to rigidity and denial of change. Yet, not only is such an inference irrational but it does not even correspond to existing judicial practice.Footnote ³ Therefore – and to hark back to H. L. A. Hart once againFootnote ⁴ – the quest for certainty is of vital importance in attempting to soften the shadows of uncertainty.

At the same time, uncertainty in terms of unpredictability, although seen as a ‘necessary evil’ by its discussants, ought to be avoided in order to strengthen state compliance with judicial rulings.Footnote ⁵ Understandably, therefore, both practitioners and scholars, albeit through different paths, have tried and are still struggling to manage uncertainty in international law in order to better apprehend the nature of CIL and how it functions.Footnote ⁶ To this end and inasmuch as uncertainty cannot be eliminated, it can be argued that ‘reduced’ uncertainty is preferable as the ‘lesser of two evils’.Footnote ⁷

Vital questions also arise regarding the way CIL develops. One such question is, how does interpretation of CIL rules functions in a complex environment with a multiplicity of actors generating ‘polycentric disputes’,Footnote ⁸ if indeed the interpretation process applies to CIL at all? And where international courts interpret CIL, what kind of change can be expected with respect to the application of CIL? Is there a ‘spontaneous’ change like a Grotian moment,Footnote ⁹ or a slow-developing process that rarely leads to change in CIL rules? And is it likely to contribute to the progressive development of CIL or to its decline?

Providing answers to these questions is of crucial importance to understanding the proper function of CIL, given that uncertainty is a fact of life, and as such characteristic of the international legal system, too. Yet, what seems to be missing from the entire discourse on uncertainty is a focus on how CIL functions. Although a number of theories have been formulated, extensively discussed, and even strongly debated,Footnote ¹⁰ there are no persuasive answers that would explain when and how changes in CIL occur, the significance of CIL inertia in resisting these changes, and the effects these changes may produce with respect to a certain time and place. In other words, the dynamics of CIL, as the driving force behind its evolution, remain essentially unexplored.Footnote ¹¹

Providing answers to these questions requires in-depth understanding not only of the structure of CIL but also of the processes that occur in and out of CIL during its operation as a social system. In this respect, the discussion on uncertainty is closely connected to the complexity of the law. When lawyers refer to the complex nature of legal problems, they are pointing out not only the density of the legislation and its lack of simplicity but also the difficulty of understanding the meaning of the law. Thus, issues of indeterminacy, uncertainty and, often, unpredictability arise when multiple meanings become available through interpretation of the law.

The present chapter is divided into four parts. Section 2 focuses on the dynamic character of CIL and what dynamic complexity entails for the system of CIL. In Section 3 we discuss how complexity science may throw light on the functioning of CIL and explore CIL’s construction as a social system. In Section 4 we demonstrate the force that sets the CIL system in motion and gives it its dynamic character: this section attempts to construct a physically based representation of the interpretation process in the CIL system. Since the present analysis regards interpretation of CIL as a process, we introduce the concept of entropy, a fundamental concept of nature, to explain the functioning of CIL as a complex social system. In Section 5, focussing on the interpretative authority of international criminal courts, we will uncover the various phases of the interpretation process followed by those courts in their case law. The analysis will, thus, adopt an entropic approach to interpretation in order to explain the progressive development of CIL.

2 The Theoretical Foundations of CIL’s Dynamic Character: A Study in Physics

It has been rightly maintained that complex laws are the inevitable price of human society that is itself complex.Footnote ¹² But what exactly is complexity in relation to CIL, and why is it something we only acknowledge ‘when we see it’?Footnote ¹³

To begin with, complexity concerns the dynamic expression of a whole – that is, a system comprising a hierarchy of multiple interacting components. As described in the natural sciences, complexity usually implies the existence of a hierarchical, multi-component system in which each component is networked and interacts with a multiplicity of other components at all hierarchical levels. Accordingly, complexity includes a process that distributes/re-distributes energy throughout the system; owing to the multiplicity of local and remote connections at all levels, this causes the system to evolve in an apparently simple but formally unpredictable manner.Footnote ¹⁴ In other words, the multiplicity of actors and the large number of issues generated between them are the main characteristics of a complex system. The prime example of such a ‘polycentric situation’ is a human society,: it involves a great number of interacting social forces that create various network structures, which in turn interact with their environment. Such a scheme is prone to change in order to adapt to new realities, especially when the existing law cannot address them.

The present analysis looks at CIL as the outcome of social processing within a constantly evolving international milieu, as originally perceived by G. Scelle.Footnote ¹⁵ It will show that this dynamic character of CIL is inherent to its social identity, since complexity serves as a frame of reference for CIL. Thus, CIL is not considered as a self-referential system, but a communicative, flexible and adaptive network analogous to physical systems.Footnote ¹⁶ In turn, interpretation, as the main function of the judiciary, will be treated as an internal property of the CIL system that initiates changes according to the particular circumstances of each individual case. In this context, interpretation plays a cardinal role in ensuring legal security.Footnote ¹⁷ Such an approach cannot of course confer on CIL the consistency and coherence it is alleged to lack, but it can explain how legal interpretation works in a complex social environment like that of CIL.

Such an approach to CIL may raise objections. Pre-emptively addressing these provides an opportunity to make a number of clarifications and further delimit the scope of the present analysis. To begin with, there is no doubt that the transfer of concepts and definitions from one discipline to another is subject to many risks. One example is the possible misinterpretation of the phenomenological basis and semantics of concepts deriving from different scientific areas. This would undermine a common effort to ask the right questions and seek the proper answers: the interpretative tools of different disciplines are by nature distinct and those of one may not fully apply, or be transposable, to the conceptual constructs of another. By way of illustration, a deterministic approach to the law does not fully correspond to how determinism functions in physics. Determinism in the natural sciences implies exact knowledge, a mathematical description of natural processes. Complicated, multiparametric physical systems involving interdependent physical processes subject to different random effects may evolve in totally unpredictable ways depending on the initial conditions.Footnote ¹⁸ Prime examples of such chaotic systems are the weather, the geomagnetic field and celestial dynamics. By contrast, the law associates determinism with strict predictability through causation; in other words, an event is causally determined if, and only if, a set of prior events has taken place. Accordingly, the law ‘argues’ that future events can be predicted with reference to prior events. The assessment of CIL is a mirror image of this construction.

If, on the other hand, we oversimplify the way things work, we tend to underestimate the complexity of both nature and society and risk producing unwarranted assumptions that do not correspond to reality – an outcome that often occurs to the detriment of both disciplines.Footnote ¹⁹ Trying to find all the answers to the questions presented is a most ambitious undertaking that cannot be realized within the restricted limits of the present chapter. Instead, our analysis will merely try to define a basis for elaborating a functional model to explain how judicial interpretation of CIL works as the main force for the motion and, accordingly, the development of CIL. The ensuing analysis will further explore the mechanics of CIL by placing a special focus on international criminal justice.

The findings of international criminal courts and tribunals are of interest mainly due to the strong ties that international criminal law maintains with CIL. As one of the sources of international criminal law, CIL has been applied not only to fill gaps in the characterization of ‘criminal conduct’ as a war crime or a crime against humanity, but also to introduce new crimes for future mass atrocities that allegedly contravene the principle of legality. In this regard, much attention has been given to the case law of the ICTY – in contrast to the ICC, whose statutory foundations in the Rome Statute has had the effect of reducing the role of CIL, since the latter constitutes a secondary source of law for the ICC.Footnote ²⁰ This is not to say that CIL has lost its importance; instead, it functions as both a source of international criminal law (ICL) and an interpretative instrument for the purpose of clarifying or filling gaps in the text of the Rome Statute, which ‘is not in all aspects an authoritative account of the current state of the law’.Footnote ²¹ For there is persistent debate over possible breaches of the legality principle,Footnote ²² which creates even more uncertainty than that which interpretation of the relevant CIL rules seeks to remove.

3 CIL as a Complex Social System: The Shared Understanding of CIL

As the outcome of ‘social constraint’,Footnote ²³ CIL is distinguished by its diversity and adaptability to a constantly changing societal environment, thus reflecting the maxim ubi societas, ibi jus. It is this dynamic character of CIL that makes it a complex system and, in particular, a complex social system.

The idea of explaining societal functions and interconnections in terms of complexity has its roots in systems theory as proposed by Talcott Parsons in the 1950s and much later by Niklas Luhmann at the beginning of the 1990s. Concurrently, system dynamics was introduced by Jay Forrester in the mid1950s as an aspect of systems theory, though in a different setting. Whereas Luhmann held that social systems, including the law, are closed ones tending to (re)produce themselves, Forrester, an engineer, introduced a methodology and a mathematical model for understanding and solving complex problems, initially in the corporate domain but later also in non-corporate areas where the focus was on the world’s socioeconomic system.Footnote ²⁴ Notwithstanding the influence of Luhmann’s theory on the law, the author of the present study argues that CIL is an open system whose interacting elements also interact with its environment.Footnote ²⁵

In short, complexity theory as developed to explain social behaviour rests upon the findings of sociology, which viewed societal institutions as autopoietic, dynamic and chaotic. However, the concept of complexity originated in the physical sciences is not defined, as sociologists or lawyers would expect, according to their normative background. As a result, the concept of complexity is often confused with the concept of complicatedness. A complicated system is one made of an intricate arrangement of many individual components and can be understood by explaining the function of each of its constituent parts. On the other hand, complexity concerns the dynamic, evolutionary behaviour of a system and is manifested by dynamically emergent properties that arise from interactions between its constituent parts; complex systems can be understood only by interpreting the multiple interactions between their multiple components.Footnote ²⁶ Besides, the exchange of energy within and through a system’s boundaries is of critical importance in comprehending the functioning of open complex systems. Accordingly, ‘complexity’ refers to the capacity of a ‘living’, dynamic system to change, adapt and evolve and should not be confused with complicatedness, which refers to a collection of discrete elements that can be de-assembled and re-assembled. Thus, a complicated problem is usually difficult to solve, although it may be simple in its structure, whereas a complex problem may be understood by studying how the whole system operates.Footnote ²⁷

Therefore, to consider CIL a complex system is to acknowledge that human society is part of nature and functions in compliance with its fundamental laws. As pointed out in Section 2, the construction of an adequate model of CIL is essential to understanding how complexity works in this area of international law. To be functional, such a model must be representative of the society it simulates or, in mathematical terms, be isomorphic to the physical setting of CIL. To clarify, isomorphic models are not identical to the entities they represent; instead, they are structured in such a way as to preserve the arrangements and binary relationships between the elements of those entities. Let us give an illustration from elementary algebra: adding together two numbers and then multiplying the sum by a third number gives the same result as multiplying each natural number by the same third number and then adding together the products (distributive property); so these sets of numbers are isomorphic, since one set can be mapped onto the other set of numbers.Footnote ²⁸ In the present analysis, isomorphism can aptly represent the two-way relationships developed between the elements of the ‘complex CIL system’. Accordingly, and for the purposes of this study, the CIL model will comprise a network of multiple, connected (interacting) states, each engaged in two-way, one-to-one interaction with international criminal tribunals.Footnote ²⁹

At this point, it is necessary to clarify that the current CIL social order is shaped by various actors, both state and non-state, operating at the international or transnational level. International and regional organizations, general or specialized, play a significant role in building the ‘practice’ endorsed by states; armed groups originating from various environments contribute to building ‘practice’, often through their involvement in conducting military operations; judicial institutions even forge state practice through compliance mechanisms.Footnote ³⁰

An illustration of the diversity of actors who join other actors in the field of CIL can be found in the recent Ongwen case before the ICC.Footnote ³¹ The ICC stressed that human communities shape their own concept of property according to their customary rules and principles. For the ICC, the content of such ‘communal property’ must be taken into account when considering the material element (actus reus) of ownership in relation to the war crime of pillaging within the context of a non-international armed conflict.Footnote ³²

Such a kaleidoscope of actors renders the system of CIL even more complex, adding components and interactions that call for further study in respect of their input and output. The diversity of actors may also lead to the emergence of multiple clusters;Footnote ³³ new relationships will arise between the agents of the system, and the feedback the agents receive from one another but also from their environment will probably lead to new processes and patterns of behaviour. This two-way feedback will reflect shared or opposing views of the law.Footnote ³⁴ Such movement entails exchange of energy between the various components of the system and may influence its evolution. Yet, one thing is certain: a complex CIL system is dynamic, since it can change in response to changing circumstances over time.

The United Kingdom’s position on air strikes in Syria in 2013 and 2018 is illustrative of the interactions between the diverse actors within the complex CIL system and between CIL and its environment, since it reveals how CIL is assessed and functions. In deciding to join with the United States and France in their efforts to reduce Syria’s chemical weapons capability, the UK Department of Defence argued in favour of a right to humanitarian intervention as the legal basis for its actions.Footnote ³⁵ It is significant to note that in 2013 the UK did not take part in the air strikes since the British parliament rejected the government’s motion for intervention. In both 2013 and 2018, however, the UK tried to establish a general doctrine of intervention applicable in situations of humanitarian necessity and to give the doctrine the force of customary law.

What is significant here is the reaction of several states to the UK’s position on the right of humanitarian intervention. Those actors and the various interactions between them and the UK form a cluster (often ephemeral) within the CIL system. Systems do not function in a vacuum, but operate within a specific environment which affects them through the exchange and redistribution of energy. In a complex legal system, like that of CIL, the flow of information regarding the meaning of the law is defined to be the energy source of the system (see Section 4).

The UK’s complex system lies within the environment of the United Nations; that environment affects the UK through the UN Security Council’s decision on air strikes against Syria.Footnote ³⁶ Although the air strikes were not qualified as an act of aggression, it seems that there was no agreement with the UK as to the customary law status of humanitarian intervention, since no state referred directly to a right of humanitarian intervention in accord with the UK’s reasoning.Footnote ³⁷ It is interesting to note that the states which rejected the proposal by the Russian Federation to condemn the air strikes – eight out of fifteen – did not openly disagree with the UK on the need to alleviate humanitarian suffering nor on the need to ensure compliance with the prohibition on the manufacture and use of chemical weapons. The absence of such feedback could signify that a shared understanding on the content of humanitarian intervention by the states was wanting.

The principal characteristic of the complex environment of CIL is the diversity of the components’ internal activity, explained by the fact that they possess a considerable degree of freedom:Footnote ³⁸ it is the discretionary power of the judiciary that confers such diversity on the agents of the system. Under the influence of Hersch Lauterpacht, the International Criminal Court underlined that the ‘constant presence of judicial discretion in matters of customary international law [entails] the “constant necessity of making a choice … between conflicting and equally legitimate principles of interpretation”’.Footnote ³⁹ In the context of an international organization (the UN in the UK example), the multiple interactions between states and other actors even accelerate the process of international standard-setting. In this respect, the UK example mirrors the model of an open system that can adapt to changing circumstances thanks to the degrees of freedom that its agents possess. In turn, when making choices, international courts cannot ignore the statements made by state leaders – nor their silence – within an international organization; thus, the courts can adapt to the developing needs of the international community.Footnote ⁴⁰

Accordingly, there may be a two-level complexity in the structure of a CIL system, with a number of complex CIL sub-systems operating within the main system (see Figures 1 and 2).

Figure 1 A one-to-one, binary relationship

Figure 2 Security Council draft resolution proposed by the Russian Federation

The result is a network, in which a broad array of interactions link the system’s various actors, the nodes of the system.Footnote ⁴¹ Therefore, understanding the binary relationships between states and international courts as a one-to-one correspondence constitutes the first crucial step in explaining the function of CIL as a complex system, and in turn its dynamic character.

4 The Mechanics of CIL: A Manifestation of Entropy

Having designed our CIL complex system with its basic units and interacting constituent parts (states, international courts, international organizations and other non-state actors as agents), we shall now proceed to explore the forces that cause it to react and stimulate its evolution. We will refer to these dynamic processes as the ‘mechanics’ of CIL and its ‘laws of motion’.Footnote ⁴² It is the relationship between the force that excites a system and the motion it causes that defines the dynamic response of the system, causing it to change, if necessary adapt and subsequently evolve.Footnote ⁴³ This is true of the CIL system, which is prone to change, to adapt whenever required and, therefore, to evolve. In this respect, the CIL is not a deterministic system. This fundamental characteristic of CIL constitutes both its strength and its weakness, since unpredictability is the offspring of change, whereas predictability – or, better, foreseeability – is an essential requirement of our legal universe.Footnote ⁴⁴

In order to account for these conflicting characteristics, we will have recourse to the concept of entropy as applied by information theory. Rooted in the field of thermodynamics and statistical mechanics, information entropy is the measure of the number of possible ways in which information is distributed in a system. A neologism derived from Greek, the term and concept of entropy was introduced by Rudolf Clausius in the mid nineteenth centuryFootnote ⁴⁵ to describe a thermodynamic property of all physical systems. To the German physicist, entropy was related to transformation and change, since thermodynamics deals with the exchange of energy between different physical systems and the work done during this process.Footnote ⁴⁶

The melting of an ice cube is the classic illustration of what entropy is, given that heat is a form of energy corresponding to a definite amount of mechanical work: when heated, ice turns into water, leading to higher internal energy (entropy) as the molecules have more freedom to move around in the liquid than in the ice cube’s solid state. For the liquid to revert back to a solid state, thermal energy must be removed and the new ice cube will not be the same as the old one. Thus changes are irreversible.

In turn, this led to the ‘second law of thermodynamics’, which is one of the most fundamental axioms of nature and essentially states that entropy cannot decrease. Later, Ludwig Boltzmann linked entropy with probability in relation to the thermal motion of particles in systems with large numbers of agents (like gases composed of large numbers of molecules), thus introducing the discipline of statistical mechanics.Footnote ⁴⁷ In other words, the second law of thermodynamics is explained as the system’s tendency towards more probable states. Consequently, entropy tends to maximize when the internal thermal energy is evenly distributed among the agents of a system – that is, when all probabilities (expectations) tend to be the same and no energy can be extracted from the system.

Information entropy offers an analogous definition: here, entropy is the measure of the amount of uncertainty represented by the probability distribution of unavailable information in a system.Footnote ⁴⁸ The entropic state of a physical or informational system cannot be reversed unless an outside force removes internal energy from the system, but that will always be at the expense of increasing entropy elsewhere and in the Universe at large. Consequently, entropy is the sum of the probabilities of all possible microstates (configurations) of a physical or informational system.Footnote ⁴⁹

As already explained, entropy defines the thermodynamic state of a physical system, which may be reversible or irreversible depending on the ability of the system to exchange energy with its environment (i.e. with other systems).Footnote ⁵⁰ In closed/isolated systems that cannot exchange energy with their environment, entropy is irreversible; although it cannot decrease, entropy can remain constant, in which case the system is said to rest in thermodynamic equilibrium.Footnote ⁵¹ In short, isolated systems are incapable of changing and, thus, evolving. If a system is not isolated and can exchange energy with its environment, then it cannot remain in equilibrium and must evolve. Human societal constructs are by nature open systems, as they are embedded in, and continuously interact with, other human societal constructs: their internal energy is subject to continuous redistribution, putting them in a state of perpetual non-equilibrium resulting in continuous change and evolution.

The definition of entropy in statistical mechanics is, to all intents and purposes, identical to that introduced by Claude Shannon in information theory/communication science in 1948.Footnote ⁵² It applies to social systems, and most certainly to legal network systems such as CIL. There, the exchange of information assumes the role of the exchange of energy in inducing excitation and response of the system. In non-technical parlance, Shannon entropy quantifies the expected value of information in a message:Footnote ⁵³ the more information is transmitted/received, the less uncertainty results and the less information entropy in the system, causing the possible outcomes of the exchange of information to be highly predictable. Flipping coins or rolling dice provide characteristic examples of measurable Shannon entropy – namely, the difference between the information contained in the message and the part of information provided by the message (i.e. the coin or the dice). Before we flip a coin, we are uncertain about what will happen when it is flipped. After the coin is flipped, the uncertainty immediately drops to zero since we now know the response (i.e. heads or tails). This, in turn, means that we have gained information on the content of the message. As a measure of uncertainty of a random variable, this definition of entropy is probabilistic in nature; accordingly, the information exchanged is a measure of that entropy’s reduction.Footnote ⁵⁴

Transposing this fundamental physical process to CIL allows us to make some very interesting observations. The CIL system’s elements – namely, states and courts – exchange information regarding the meaning of CIL; states respond through their practice and courts by interpreting rules. The greater the number of probable meanings assigned to the rules in question, the greater the entropy of the CIL system; conversely, when there are fewer probable meanings, there is less uncertainty as to the content of the rule and thus minimal entropy. To this end, legal interpretation plays a fundamental role in reducing the entropy of the CIL system by providing the information that was originally not available (i.e. the concrete meaning of the CIL rule). In particular, the exchange of information about the most probable, or even precise, content of a CIL rule creates a network of binary relations between the courts that transmit the message contained in the information and the states that receive it. When international courts like the ICC form binary relations with states, this creates a one-to-one connection between them, since they are the sole components of the complex CIL system concerned.Footnote ⁵⁵ The transmission of the precise content of a CIL rule to states by international courts through the interpretation process shapes the connections between the courts and the states, for courts receive feedback from the states, regardless of whether the states comply with the court’s judgment.

Having said this, the analysis will now focus in Section 5 on what we call an entropic approach to the interpretation of CIL and how it is reflected in the jurisprudence of the ICC and other international criminal tribunals.

5 Modelling the Interpretation of CIL: What Do International (Criminal) Courts Say?

Traditionally, the function of the judicial power – that is, courts and tribunals stricto sensu – is to state the law (i.e. not only to decide which legal rules are applicable but also to determine their content in a concrete case).Footnote ⁵⁶ The legal reasoning that unfolds during the process of interpreting a CIL rule has a concretizing function (i.e. to give a concrete meaning to a CIL rule, the content of which is by definition general).Footnote ⁵⁷ Legal interpretation is an exercise akin to the solving of a mathematical problem; when judges seek to determine the content of a rule in a particular case, they apply their skills to solve the problem by either settling the dispute or establishing the guilt of an accused individual.

In the case of CIL interpretation, courts determine the content of a rule by searching for states’ shared understanding of the rule. Such was the approach taken by the International Criminal Court when its Appeals Chamber dealt with the thorny issue of immunity for heads of state having committed international crimes under the Rome Statute in Sudan. As Judges Eboe-Osuji, Morrison, Hofmanski and Bossa in their concurring opinion openly stated, ‘the consistent and repeated rejection of immunity (even for Heads of State) in sundry instruments of international law since World War II … has resulted in a general understanding of customary international law in that way’.Footnote ⁵⁸

Yet, interpreting CIL rules in a criminal law context runs the grave risk of overstepping the principle of legality, which is one of the basic pillars of international criminal justice. Since international criminal courts are not permitted to ‘expand uncritically the scope of [international] crimes’,Footnote ⁵⁹ they are confronted with a challenge when called upon to determine the content of a CIL rule with a view to characterizing it as a core international crime.

In the Kupreskic case, the ICTY adopted an interpretation with interesting reasoning when, following extensive analysis, it defined the content of the residual category of ‘inhumane acts’ as crimes against humanity.Footnote ⁶⁰ Based on the notion of ‘humane treatment’ included in Common Article 3 of the Geneva Conventions 1949 and international human rights standards, the ICTY ruled that it is possible to ‘identify a set of basic human rights’ whose violation could amount to a crime against humanity. Bearing in mind that Common Article 3 had already been recognized as having erga omnes effect and that basic rules protecting human rights have the force of customary law, the ICTY defined the notion of inhumane act as a rule having ‘less broad’ customary law content.Footnote ⁶¹ The flexibility with which the ICTY handled the issue of CIL interpretation is particularly revealing, albeit not always plain to see. The case of Krstic is also indicative, since the ICTY recognized the concretizing function of CIL, which had delimited the actus reus of the crime of genocide so as to include only acts of physical destruction, a limitation accepted both by the ICTY and the ICTR in their respective statutes.Footnote ⁶²

Most significantly, the Appeals Chamber of the International Criminal Court followed the same path in one of its most recent judgments in the Ntaganda case,Footnote ⁶³ although not in order to directly downsize the CIL rule; instead, the its aim was to ensure consistency with international humanitarian law (IHL), in both its customary and its treaty forms. Such an outcome can be achieved either by interpretation of its terms or by identifying additional elements to the crime in question. In particular, the Appeals Chamber came to an ‘unprecedented’ conclusion when it accepted that additional elements originating from CIL can be included as an element of the war crimes listed in Article 8(2)(b) or (e) of the Rome Statute as part of examining the ‘established framework of international law’.Footnote ⁶⁴ Although not openly stated, it is clear from the Court’s analysis that if an element external to the Rome Statute is added to the war crimes enumerated as other serious violations of the law of war, such an element can and will be interpreted, irrespective of whether it derives from customary or treaty IHL. Besides, the Court explicitly ruled that ‘the “established framework of international law” within article 8 (2) (b) and (e) of the Statute permits recourse to customary and conventional international law regardless of whether any lacuna exists.’Footnote ⁶⁵ In short, the ICC recognized the interpretability of CIL and rejected allegations of possible violation of the legality principle.Footnote ⁶⁶ This is a novel approach, not only for ICL but for the function of the judiciary in general. As successfully argued by Scelle back in 1948, jurisprudence ‘assouplit et adapte’ (softens and adjusts) legal rules in accordance with evolving social needs. Since the judicial function grows out of the social environment through natural evolution, its technique has a dynamic character. In this respect, jurisprudence, fulfilling the law, is part of the judicial order.Footnote ⁶⁷ The interpretation process as just described allows CIL, thanks to its equally dynamic character, to adjust to changes in its social environment. The reactions of the acting agents in the CIL system (i.e. states, non-state actors and international criminal courts) provide feedback into the system, driving it to change and evolve. Entropy is thus at a low level, since uncertainty over the content of the CIL rule has been reduced.

This dynamic process was analysed by Judge Antonio Cassese when, as president of the Special Tribunal for Lebanon (STL), he delivered an interlocutory decision on the applicable law in the Ayyash case.Footnote ⁶⁸ In explaining the role of the interpreting judge, Cassese recalled the constructive function of interpretation as being ‘to give consistency, homogeneity and due weighting to the different elements of a diverging or heterogeneous set of provisions’.Footnote ⁶⁹ This approach to interpretation is founded upon the context of the rule, both internal and external. In particular, it was stressed that ‘the rules of interpretation that evolved in international custom and were codified or developed in the 1969 Vienna Convention on the Law of Treaties … must … be held to be applicable to any internationally binding instrument, whatever its normative source’.Footnote ⁷⁰

As CIL is unwritten law, it falls upon the courts to elucidate relevant rules in their context – in other words, to contextualize the CIL rules.Footnote ⁷¹ Thereby, CIL can be interpreted according to the general principles of judicial interpretation with a view to ‘apply[ing] legal norms consistently’.Footnote ⁷² The constructive approach, adopted by the STL, is inextricably linked to teleological interpretation. Taking the form of the effectiveness principle, contextual interpretation aims to ‘harmonise the various provisions in light of the goal pursued by the legislature’.Footnote ⁷³ In giving concrete meaning to a CIL rule, it reconciles the various meanings of the CIL rule. Subsequently, entropy is reduced, as harmonization counteracts uncertainty by revealing the most probable meaning of a CIL rule. In this respect, a CIL rule will consequently reflect the shared understanding of international actors regarding the content of the rule.

For the STL, such an interpretive approach corresponds to the reality of a dynamic society that changes over time, thus leading to the progressive development of the law. Recalling Jeremy Bentham, the presiding judge, the late Antonio Cassese, clarified that CIL also develops through a ‘continuing and prospective evolution’ in order ‘to adjust [the Court’s] decision to the special circumstances [of a particular case]’.Footnote ⁷⁴ Progressive development does not preclude the application of the legality principle, ‘as long as this application was foreseeable’.Footnote ⁷⁵ To the extent that international actors – state and non-state – are affected by that interpretation, they will react by providing feedback to the Court’s interpretative approach.Footnote ⁷⁶

At this point, some clarification is needed: obtaining feedback from states in ICL cases is not of the same nature as in other international law interstate cases such as those before the ICJ. Feedback from states in ICL cases involving CIL interpretation can assume different shapes and forms. Take, for example, Articles 86–87 and 97 of the Rome State on the obligation of state parties to cooperate and consult with the ICC during the investigation, prosecution and enforcement stages. Here, the triggering of a mechanism of non-compliance with the Court’s judgments constitutes proof of feedback by a member state. Let us elaborate on the facts and the Court’s reasoning to comprehend how entropy changes in this case.

The judgments against South Africa and Jordan, parties to the Rome Statute, concerning non-execution of the arrest warrant against Omar al-Bashir, are examples of such a complex case. The ICC ruled that both states had violated their obligations deriving from the Rome Statute, since they had not arrested and surrendered Sudan’s head of state al-Bashir to the Court for having committed war crimes, crimes against humanity and genocide.Footnote ⁷⁷ Pursuant to a UN Security Council referral of the situation in Darfur to the ICC, the latter issued two arrest warrants for al-Bashir in 2009 and 2010 calling upon all state parties to cooperate with the Court in accordance with their obligations under the Rome Statute.Footnote ⁷⁸ After rejecting the objections of South Africa and Jordan invoking the immunity rule for heads of state, the Court explained that there was no CIL norm on immunity that shielded active heads of state accused of having committed core international crimes.Footnote ⁷⁹ The most interesting conclusion of the ICC Pre-Trial Chamber was its decision to refer the non-compliance issue to the Assembly of States Parties and the UN Security Council. However, it did so only for Jordan, not for South Africa, reasoning that the former was aware of its obligations but ‘chose not to execute [them]’,Footnote ⁸⁰ whereas the latter was not. The second Pre-Trial Chamber (PTC II) maintained a different approach towards South Africa based on two main reasons: firstly, there was genuine interaction with South African authorities regarding the state’s obligation to co-operate with the Court and, secondly, South Africa’s violation of its obligations had also been recognized by the country’s domestic courts.Footnote ⁸¹ The Appeals Chamber overturned the decision for Jordan and decided not to refer the situation to the Assembly of States Parties and the UN Security Council. It held that Jordan had not refused to consult with the court in defiance of its obligation deriving from Article 97 of the Rome Statute. On the contrary, its note verbale qualified as a request for consultations with the Court and was proof of good faith with regard to the problem of the execution of the arrest warrant.Footnote ⁸²

We draw particular attention to the feedback of the Court based on the different legal reasoning of the PTC II and the Appeals Chamber. Although not directly connected to interpreting CIL, each legal reasoning is important insofar as it portrays how interpretation works as a means of keeping the entropy of the CIL system at a low level. Further evidence of that is the Appeal Chamber judgment concluding that the different treatment of Jordan and South Africa was due to abuse by PTC II of its discretionary power. When state parties rejected the Court’s interpretation regarding their obligations under the customary rule of immunity, the entropy of the system increased, as there were many possible interpretive (and conflicting) outcomes. When the Appeals Chamber overturned the PTC II’s judgment, the CIL entropy had been reduced to a minimum as there was only one possible interpretive outcome for the CIL rule on immunity. This shows that the process of ICC feedback to states proves to be particularly influential for complex systems, since it keeps the entropy at a low level.

International case law can also affect state behaviour by triggering a change in state practice.Footnote ⁸³ The feedback process leads the CIL system to adapt through a learning process that relies on information from its own agents as well as its environment. In other words, the topology of the CIL system (i.e. the structure of the CIL network and the frequency and intensity of interactions between its agents) is indicative of its inherently dynamic character and its ability to evolve as a real social phenomenon.

6 Concluding Remarks: A ‘Low Entropy’ Mode of Interpretation for CIL

The present chapter has attempted to show how the interpretation of CIL operates in accordance with the fundamental rules of nature. Since human systems are affected by various kinds of complexity, we turned to the findings of complexity science to describe CIL as a complex adaptive system forming societal bonds between the various state and non-state agents involved in international criminal justice.

In modelling CIL as an epistemic tool, we have sought to explain the mechanical behaviour of CIL when international criminal tribunals and, in particular, the ICC interpret its rules. To this end, we introduced entropy, as the internal property of a complex system that stimulates changes according to the flow of information transmitted both between its component elements and between the latter and the system’s environment.

For the CIL complex system, legal interpretation is the source of information provided by the international courts when they concretize abstract CIL rules by attributing the most probable meaning to those rules. Such an entropic mode of interpretation reduces uncertainties over the content of the rules, triggers changes in the CIL system and, thus, encourages the system to develop further and thereby evolve. The extent of the changes that will occur depends on the feedback such interpretation draws from the agents of the CIL system.

Adopting such an approach to the interpretation of CIL helps us redefine our perceptions as lawyers regarding the quest for certainty and predictability in international law. Thus, we have argued that chance does not correspond to the unknown and the uncertain that Czech choreographer J. Kylian considered to be epitomized in the tossing of the dice; rather, throwing a dice is all about probabilities.

In the field of international criminal justice, reducing uncertainties is especially vital to the unity and progressive development of international criminal law. Recent interpretation by the ICC has displayed low entropy. Were the Court to follow such an entropic interpretation in the future, this could lead to the transformation of CIL and in turn to a unified approach in ICL. It remains to be seen whether the international judiciary will seize the opportunity to ‘activate’ entropic interpretation to stimulate changes in the complex CIL system.