This article examines the role of international law in a risk society. The association between the two is fairly intuitive as, by definition, risk respects no state borders; it is perhaps one of the few phenomena that can be defined as ‘global’ in and of itself.Footnote 1 It thus seems reasonable that, being itself concerned with global issues, international law would serve as a framework for tackling the challenges of a rapidly evolving risk society. Ulrich Beck, one of the leading theorists of the risk society, seems to think as much. When discussing the risk of global terrorism, Beck imposes a heavy burden on international law:
[I]n a nationalist context, that which infringes upon the legal sensibilities of the civilized world is the fact that the victims of the attempts assume the role of persecutor, judge and executive power at once. This type of ‘self justice’ must also be overcome in international relations. Even if relations between the states are not fully ripe for it, the global alliance against terrorism has to be based on the law. Thus it follows that an international convention against terrorism must be discussed and ratified.Footnote 2
The same argument could be made with regard to certain financial risks, health hazards, or pollution. Risk of global terrorism epitomizes the general uncertainty inspired by an unidentified enemy that is always there, ready to harm. In this context international law is called upon as an independently normative force, in the hope that it brings some order to a seemingly chaotic situation. Such is Beck's point when he calls for further international regulation as a positive force within the risk society. Now, is international law up to the task? I argue that it is not. International law is unable to serve as an independently normative force of the risk society, as it lacks the tools to deal with the cornerstone principles of such a society: randomness and chance. International legal reasoning has a hard time dealing with purely fortuitous events, which constitute the most characteristic phenomena of the risk society. As a result, when faced with random events, international law seems to be forced to outsource its normative authority, referring to non-legal expertise in order to give acceptable solutions to concrete cases. International law thus becomes a broker of expertise, which refers to the relevant epistemic community that is needed to ‘get the job done’ in risk societies, and abandons any aspiration of holding an independent normative pull.
In what follows I shall detail the reasoning that leads to such a conclusion. Much of my discussion here is focused on the notion of causation, both in general legal reasoning and in international law. This is so because causation is the only prism provided by law through which randomness can be observed. Random events are also events for which we are unable to provide a causal explanation; therefore, mere chance is also non-causation, and can be usefully studied as such. The prism, however, is broken. Borrowing Dworkin's celebrated image, this article argues that causation is doughnut-shaped, featuring in its centre a wholly unregulated process (the hole) where assessment of causation is ultimately done. There are no distinctively legal criteria to identify a cause: in effect, legal causation requires us to appeal to non-legal variables in order to determine causation. When international law faces randomness in risk societies, the same problem applies: international law has to appeal to non-legal variables, thus becoming in effect a mere broker of expert knowledge, with the consequences announced before. This is not to say that international law is alone in its inability to frame randomness and chance. Domestic law is also unable to do so, as such limitations derive from the very notion of legal causation that is shared by the two systems. However, in the context of the global risk society, the consequences for each legal order differ: while domestic law's independent normative pull is not questioned as a consequence of these limitations, international law's is. And this, in turn, has worrying effects on the possibility of achieving something along the lines of an international rule of law. We shall discuss this soon. Now it seems fitting that we consider risk societies in some detail, and their link with randomness and chance.
1. Risk society and randomness
Risk is Janus-faced, and randomness is its correlate aspect. This section explores some fundamental features of Beck's risk society from the perspective of randomness and chance. As we now turn to see, Beck's manufactured risk effectively underscores the role of randomness in contemporary society. This move, as will be discussed in the next section, proves to be quite problematic for the relevance of international law today.
Let us consider Jérôme Kerviel. Kerviel allegedly managed to cause a €4.9 billion write-down at the Société Général (SocGen) in 2008.Footnote 3 He was not the first. In 1995, Nick Leeson caused the collapse of the Barings Bank with a US$1.8 billion loss,Footnote 4 and, more recently, a Canadian, Brian Hunter, brought down Amaranth Advisors with a misplaced bet on gas futures of US$6.5 billion in 2006.Footnote 5 Regardless of whether Kerviel's case actually turns out to be one of such debacles, we all seem to agree that this is quite possible: it does seem possible that a mid-level trader's mischief hits the market with a loss in excess of Mongolia's annual gross domestic product (GDP).Footnote 6 It is more than possible; it is really not surprising at all – that is, after all, the world we live in.
But is it really? These men are derivatives traders who specialize in complex financial instruments. Risk is their trade. Their stories of rise and fall seem closer to the world of Gordon Gekko in Oliver Stone's Wall Street than to our world, of home mortgages and credit card payments. However, as SocGen's €4.9 billion hit was sending shock waves through the global financial system, we all seemed to be involved. When the time comes, we all inhabit their world, a world rightly described by Beck as a risk society.Footnote 7
According to Beck we are living a new modernity. The first modernity transformed the planet in a deep, essential sense, to an extent never before witnessed. The depth and scope of such transformation implied, in turn, the emergence of numerous hazards that are unpredictable from a human perspective. Risk is our effort to deal with such an unpredictable scenario; in Beck's own words, risk is ‘a systemic way of dealing with hazards and insecurities induced and introduced by modernization itself’.Footnote 8 The second modernity is the era of risk: whereas the first modernity was concerned with wealth creation and distribution, the second modernity is concerned with risk aversion and distribution.Footnote 9 The risk society is, as Giddens puts it, ‘a society where we increasingly live on a high technological frontier which absolutely no one completely understands and which generates a diversity of possible futures’.Footnote 10
The move to a new modernity is performed through ‘reflexive modernization’. There are two versions of such a reflexive process: one offered by Giddens,Footnote 11 a second by Beck.Footnote 12 Giddens's account focuses on the idea of reflection. In his view, modernity reflects upon itself and modernizes even further, thus creating a new modernity. He is ultimately concerned with the reflexive citizen: someone who holds certain knowledge on the effects of modernization and reflects (thinks) on them, thus transforming modernization. Beck, in turn, accepts Giddens's account, but argues that reflexive modernization is a much narrower concept.Footnote 13 In his view, reflexive modernization does not concern reflection, but reflex; that is, ‘the effect or preventive effect of non-knowing’.Footnote 14 The key concept of the new modernity is, then, not knowledge, but unawareness.
Risk is an important element of this equation. In a risk society, we face complete uncertainty concerning the unintended consequences of the modernization process. There are hazards out there of which we are wholly unaware. Beck's recurring example of radioactivity comes to mind: we cannot see it and we do not know exactly how it affects us.Footnote 15 But we know it is out there. Hence we react reflexibly – not by reflecting on the issue (as Giddens would argue) but by having the political equivalent of a knee-jerk.
This is not to say that risks are exclusively the offspring of new modernization. Beck is quick to recognize as much:Footnote 16 trees have been dying for a long time now, and certainly the 1929 crash featured a more worrying landscape than the mere crisis of a hedge fund.Footnote 17 The difference lies in the utter uncertainty we are facing today, and its sources. We are so amazingly unaware of the hazards that may come, that such hazards become incalculable. Hazards are not simply unknown; they are unknowable through our own experience, and our reaction to such uncertainty constitutes the risk society.
Perhaps the best way to understand this qualitative leap is by comparing it to a ‘community of risk’.Footnote 18 A community of risk is present when potential injurers are also potential victims, and equally so. That is the case of a high-speed highway: all drivers are at the same time potential injurers and potential victims. In the example, even though hazard is omnipresent and crucially important, it is still calculable and subject to knowledge. When driving, although in perennial danger, we know where harm may lie (other drivers, wet pavement, low visibility, etc.) and are not concerned with hazards outside that scope. Regardless of whether we are victim or injurer, we are able to imagine the contours of an event and its plausible consequences: car damages, injuries, death, and so on. Moreover, it is possible to calculate the probability of actual harm rising from those hazards (hence the insurance industry), and it is also possible to take regulatory action to decrease such a likelihood (for example, by reducing the speed limit). Plainly said, risks on the highway are known, observable, calculable, and, eventually, reducible. Such kinds of risk are not new, and may be identified with the notion of ‘external risks’.Footnote 19
Now compare that community of risk with Beck's risk society. In a risk society, everyone is both cause and effect of risks that emerge from modernization.Footnote 20 However, Beck argues that it is not possible to calculate such risks: ‘along with the growing capacity of technical options grows the incalculability of their consequences’.Footnote 21 Such impossibility of calculation hinders our very ability to know the danger that lies ahead. True enough, risks are nothing new. But the risk society is, because it concerns not external risks, but risks created by modernization itself. Such hazardous side effects of modernization (also called ‘manufactured risks’Footnote 22) are fundamentally unknown, unobservable, and incalculable. Therein lies the qualitative leap implied in the notion of the risk society.
Manufactured risk features one further dimension of importance: its subjectivity. The issue is not only that humanity as a whole may be unaware of, say, the risks of genetically modified organisms (GMOs). It is also that, as we have seen, manufactured risks are unknowable in any significant way. They remain invisible and ‘thus initially only exist in terms of (scientific or anti-scientific) knowledge about them’.Footnote 23 Thus manufactured risks are such in two different senses: first, they are manufactured as they are the product of man's intervention in the planet, and, second (and perhaps more importantly), they are manufactured as they only exist in and through human knowledge – manufactured risks are thus ontologically subjective.Footnote 24 Without human knowledge, there is no ‘risk’, but mere hazardous events.Footnote 25 However, knowledge is unevenly distributed, making awareness of risk also unevenly distributed. The very existence of manufactured risk is affected by such dynamics; manufactured risk does not exist in the same way for all humanity, for all social groups, and for all individuals within those groups. A manufactured risk is, by definition, not the same risk for everyone. There is no given stock of manufactured risks that can be found somewhere and then be shared by all humanity, for risk exists only in knowledge and interpretation. Manufactured risk is indeed subjective. The risk society is, ultimately, the exchange system of subjective interpretations of manufactured risk.Footnote 26
Smoking provides a good example. Mass production and advertising of cigarettes only began in the 1890s, when James Buchanan Duke founded the American Tobacco Company,Footnote 27 making it possible to read cigarette-related risks as a side effect of modernization. During the first half of the twentieth century, heavy smoking was not perceived to be a health hazard, quite the contrary – think, for instance, of Humphrey Bogart in Casablanca. Since the 1960s the link between smoking and disease was progressively made, first by scientists and eventually by the mass media. Today the link is generally accepted, and there is awareness of such risk. Consider, however, the subjectivity of such awareness. Think of the time gap between scientific awareness of the health risk derived from smoking and its popular diffusion.Footnote 28 Was humanity aware of the risks of smoking in the early 1970s? The difficulty in answering that question goes to show the subjectivity of manufactured risk. Think also of the difference between smoking trends in First and Third World countries.Footnote 29 Developing countries' share of global cigarette consumption has risen from 40 per cent in 1970 to 70 per cent in 2004;Footnote 30 moreover, smoking is more common among the less educated population within developing countries.Footnote 31 Regardless of what the reasons for this situation may be,Footnote 32 it does show that manufactured risks do not exist uniformly throughout the globe, as they are manufactured through knowledge, and that knowledge is not uniformly spread.
Manufactured risk is thus both unperceivable and exists only in terms of available knowledge. These two traits are not contradictory – there is no tension between them. Rather, they reinforce each other in creating the general sense of uncertainty that characterizes the risk society as a whole. Indeed, the existence of manufactured risk depends upon who perceives it, and in which context. What we call the risk society is in fact an informal network of risk perceptions, with no ultimate instance of decision. There is no global clearing house of perceptions that will tell us what the ‘true’ risks on this planet are. Such a decentralized system has important effects in our perception of events, for the chain of unintended consequences is unknown to such an extreme degree that, in effect, we turn to perceive those consequences as the product of mere chance. In this context it is ultimately irrelevant whether events are as a matter of fact random or not. The important point is that we are unaware of the risks that lie ahead, and that we are ultimately alone in our unawareness. Thus, if unaware of risk, the negative side effects of modernization will seem as mere random occurrences. The impossibility of imagining the danger that lies ahead makes us think that, when such danger materializes, it did so out of mere chance. This is an era of manufactured uncertainty which, in Beck's words,
means that risk has become an inescapable part of our lives and everybody is facing unknown and barely calculable risks. Risk becomes a word for ‘nobody knows’. We no longer choose to take risks; we have them thrust upon us. We are living on a ledge – in a random risk society, from which nobody can escape. Our society has become riddled with random risks.Footnote 33
Randomness has always been intimately linked to the notion of risk. This link, however, depends on the type of risk with which one is dealing. For instance, the relation of external risk to chance is, in essence, one of control. No sane risk manager would tell a client that ‘whatever has to happen, will happen’. On the contrary, the risk industry is based on the idea that apparent randomness is not mere chance, but actually follows certain laws of probability that can be learned and applied.Footnote 34 Manufactured risk, in turn, features a wholly different relation with chance: its essentially unknowable dimension is not intended to control chance, but brings randomness to the forefront as the most likely explanation for our perception of consequences. Randomness is, thus, the key word in the risk society.
2. Randomness and causation in legal reasoning
How does legal language frame the idea of randomness? In principle, it does not. The idea that events may be purely fortuitous seems to shock the legal mindset, and not without reason, as law is not well equipped to tackle the issue. This section proposes that causation is the only means through which we are able to make sense of the role played by randomness in legal reasoning. However, while causation has been systematically addressed in legal scholarship, absence of cause has not been considered at all. As a consequence, we lack a legal theory of randomness, a fact that deeply undermines the relevance of law (and international law, in particular) as an independent normative force within risk societies. Indeed, legal language seems to require that we look beyond law for an answer as to what a cause is and thus what a random event is. Therefore, when faced with the problem of causation and randomness, law can only give way to other languages, and yield to such languages' own political agendas. International law, it will be shown in the next section, is no exception, as it features no specifically international notion of causation, thus reproducing the limitation of its domestic counterpart.
The link between legal reasoning and randomness is based upon a fairly intuitive notion of justice, whereby legal outcomes should flow from morally relevant events. This intuition has been called the ‘control principle’, which can be traced back to the Kantian idea that morality is immune from luck, for we are morally assessable only for events under our control.Footnote 35 This principle's corollary is, of course, that we are not morally assessable for events that are outside our control.Footnote 36 The underlying concept here is voluntary action: a moral agent that acts voluntarily (that is, that could have acted otherwise) should be held legally responsible for her actions, but should not be held responsible for events that are outside her control and are thus a matter of luck.Footnote 37 It follows naturally that acts deserving moral reproach should have legal consequences, while it feels unfair that legal consequences (especially sanctions) should follow from merely fortuitous events.Footnote 38
Indifference is, then, the standard approach of legal reasoning to chance. The problem is that, despite our first intuition, we are commonly held legally responsible for outcomes that depend on mere luck. Consider the problem of attempted crime. As Hart has noted, there is ‘the almost universal practice of legal systems of fixing a more severe punishment for the completed crime than for the mere attempt’.Footnote 39 However, attempted crime differs from the completed one only in a stroke of luck – good or bad, depending on where one is standing, but ultimately luck, as the moral intention is precisely the same. Why should the lucky defendant of an attempted crime receive a lesser punishment than her unlucky counterpart? There seems to be no rational explanation and yet the difference in punishment does exist in most legal systems.Footnote 40 Moreover, we actually feel it fair that such a difference exists: in this case, it somehow feels correct that the moral assessment implicit in the criminal conviction should be based upon the sheer luck of the crime's completion. That is, in essence, the problem of moral luck: it feels correct to attribute moral consequences to mere random events.Footnote 41 It is important, however, to note that this is not a mistake. It is not that we are simply incorrect to create a differentiated treatment for attempted murder – or, for that matter, for that lady in high heels who stepped on our toes after being pushed herself. Moral luck is one important paradox of the whole process of moral assessment; it is, indeed, ‘a perception of one of the ways in which the intuitively acceptable conditions of moral judgment threaten to undermine it all’.Footnote 42
Moral luck has had important implications for debates on criminal punishment,Footnote 43 tort law,Footnote 44 and distributive justice.Footnote 45 I want to focus here on the limitations of legal language it reveals. Even though legal language is formally indifferent to luck and random events, the moral luck debate shows that such language is effectively burdened by constant considerations of chance.Footnote 46 However, because such influence is not recognized, it becomes impossible to discount explicitly the effects of randomness in any given legal decision. Luck clearly exists in legal reasoning, yet it is bound to remain hidden. There is simply no way to bring randomness to the forefront of the legal discourse. Why is this so? Limitations to considering the role of randomness seem to derive from the notion of causation implicit in legal reasoning. Causation is used as a criterion to draw the line between random and non-random events. If an event is not causally linked to another, then it is random, and, as such, is perceived as being beyond any significant legal assessment. There is, however, little explanation in legal language as to why a condition can be seen as the cause for an event, making it necessary to appeal to extra-legal criteria in order to identify a causal link and to identify a random event. It is the very notion of causation that hinders legal language from explicitly considering issues of randomness, as we now turn to see.
2.1. Legal causation
Originally, the problem of causation was not of a legal nature. The notion that ‘nothing comes from nothing’ has been troubling philosophers since ancient Greece, creating one of the longest-lasting discussions in social science. The main character of this story is the empiricist David Hume, whose philosophy of causation has set the basic assumptions around which the discussion gravitates.Footnote 47 As a good empiricist, Hume argued that causation could not be found beyond our actual perception of events. But we cannot actually perceive causation: we merely perceive constant conjunctions of events, which we then end up associating, thus creating the illusion of causation. Therefore, according to Hume, causation ‘belongs entirely to the soul which considers the union of two or more objects in all past instances’.Footnote 48 Such an approach has been deemed unhelpful for lawyers facing the problem of legal causation.Footnote 49 The Humean approach seems to be focused on two questions: (i) is there truth in the principle according to which all events have a cause? and (ii) if two events are related as cause and effect, does the cause necessarily have that effect?Footnote 50 The lawyer, on the contrary, is not concerned with those questions, but rather asks: under which conditions and for what purposes can something be considered as a cause?Footnote 51 The legal approach, then, seeks not to establish whether a general rule of causation can be extracted from the observation of several conjunctions of events, but to determine how a general principle of causation can be applied to a specific case where two events seem to be related.Footnote 52
The notion of legal causation is, thus, surprisingly narrow. Just how narrow it is can be gleaned from the tension behind international criminal trials, such as Nuremberg or the Milošević process. As legal procedures, these trials are expected to apply essential elements of legal reasoning, such as the right of contradiction and, to be sure, legal causation. However, they are also (and perhaps as importantly) ex post facto evaluations of large historical events, which are thus commonly justified on some combination of the quest for historical truth and deterrence.Footnote 53 The tension then becomes apparent. International criminal trials have an undeniable didactic aspect, which is impossible to achieve if one is focused on the narrow effort of producing evidence of a plausible chain of causation leading to the Srebrenica massacre.Footnote 54 The narrowness of legal causation implies a recurrent tension between ‘the need of persuasive storytelling and the normative requirements of liberal judgment’.Footnote 55 History, even history on trial, requires wider notions of causation than those offered by the law.Footnote 56
Legal causation is traditionally linked to the sine qua non (‘but for’) test.Footnote 57 This test is fairly straightforward in its application, and states that two events are causally related if one would not have happened without the other. The idea, then, is that a sine qua non condition of an event is ultimately its cause; hence the legal notion of causation is reduced to asking ourselves whether harm would have happened were it not for the conduct that is being assessed. If harm would have occurred absent the conduct, then the latter is not a cause of the former, and vice versa.Footnote 58 This notion defines, in turn, the issue of randomness in law: if the event (for example, harm) is not causally linked to the conduct that is assessed, then such harm is the product of chance with regards to that conduct, thus becoming irrelevant for the assessment of said conduct. Consider, for instance, a car crash due to a driver not stopping at a red light: under the ‘but for’ test, injuries are causally linked to the conduct because they would not have happened had the driver stopped at the light. However, if injuries are not causally linked to the crash (if they are, say, previous wounds), then they become a mere random event with regard to the driver's conduct and are thus irrelevant for the assessment of the conduct itself – or, rather, are relevant only inasmuch as they are considered the product of mere chance. In this sense, causation and chance are but two sides of the same coin.
Despite its relatively narrow aspirations, legal causation raises two different kinds of objection: (i) the very notion of ‘legal causation’ is objected to as a contradiction in terms; and (ii) legal causation as a mere ‘but for’ test is also resisted. The present argument on randomness, however, requires that we focus only on the first block of objections, for it is the very notion of legal causality (and not the specific test whereby it is established) that hinders legal language from considering explicitly issues of randomness.Footnote 59 For that purpose, it is useful to focus on Hans Kelsen, who famously worded the first objection. Kelsen draws a clear line between the is and the ought, arguing that such a line cannot be crossed, for the two modi are fundamentally different.Footnote 60 In his words, ‘nobody can state that from the statement that something is follows a statement that something ought to be, or vice versa’.Footnote 61 Law is, of course, the realm of the ought; by contrast, Hume observing billiard balls that create conjunctions of events is the realm of the is. One cannot jump from one to the other without fault; in this way the notion of legal causation would be fundamentally flawed. To make his point, Kelsen opposes the notion of causality to that of imputation: causality exists in natural sciences and imputation in law.Footnote 62 Therefore, in his words, ‘the principle of causality states: If there is A, there is (or will be) B. The principle of imputation states: If there is A, there ought to be B’.Footnote 63 And, in Kelsen's view, what ought to be is reward or punishment, for imputation – and thus, law – is not concerned with who committed a certain act (that being a question of mere fact), but with the assessment of responsibility: who should be rewarded, who should be punished, for this act?Footnote 64
Kelsen's argument seeks to prove that legal causation is absurd, for law can only focus on imputation. However, that same argument can be used to reveal the hidden complexity of legal causation, as Kelsen's imputation can be also read as the second phase of a two-level assessment of legal causation. Indeed, borrowing Dworkin's celebrated analogy,Footnote 65 legal causation is like a doughnut, which actually involves two different level of decision-making: first, an outer level that features the highly regulated sine qua non test proper, where an assessment of all conditions for a given effect is undertaken. And, right in the middle of such a test, there is a hole: a second level, beyond the ‘but for’ test, wherein one of the conditions is singled out and chosen as the cause of the event. This second level is a hole because it is an unregulated process, as it is ultimately dependent on the context of the person or institution making the assessment, which defines, in turn, whether an event is random (or not). Think of the man who throws a cigarette butt in the rubbish bin and, soon after, a fire burns down the place. We would easily identify the cigarette as the cause of the fire. However, several other conditions were required for the fire to come into existence, all as necessary as the cigarette butt. Oxygen was required, flammable materials, and so on. And yet we pick the cigarette butt as the cause, making a clear normative choice of causation among several available conditions.Footnote 66 The problem is that Kelsen's approach to causation seems to imply that one effect is clearly related to one single cause, thus requiring no normative assessment when applying the so-called ‘principle of causality’.Footnote 67 Yet this is not an accurate description of reality. Hardly any effect is the result of one single condition. On the contrary, as John Stuart Mill has appropriately put it, ‘it is seldom ever between a consequent and the single antecedent that this invariable sequence subsists. It is usually between a consequent and the sum of several antecedents, the concurrence of them all being requisite to produce, that is to be certain of being followed by the consequent’.Footnote 68 There are thus causes and there are conditions. Several conditions exist jointly in order to make an event possible; however, we single out some of those conditions and call them a cause.Footnote 69 Such is the normative assessment that is performed in the hole of the doughnut, at the second level of the legal causation test. Against Kelsen's view, legal causation is indeed a viable assessment; however, it does involve an assessment beyond the mere ‘but for’ test, wherein a cause is selected among several conditions. Causation is, in this sense, also a problem of ought.
With this in mind, let us return to the problem of randomness. The normative choice that is implicit in identifying a cause is also instrumental to drawing the line between random and non-random events. When we single out a cause among several conditions, we are arguing that the event is not random, as it is causally linked to the condition we have singled out. If the event is not related to the condition we have singled out as cause, then we argue that the event was the product of mere chance. Reconsider the example of the fire. Once again, all conditions are present: oxygen, flammable materials, and so on. Once again, our man threw his cigarette butt into the rubbish bin. However, in this case lightning struck the building and it burned down. If we are not able to link our man's conduct causally to the fire (even if said conduct remains unchanged) then we would say that the fire was an event of mere chance with regards to his actions. ‘Not quite so’, a keen reader would argue, following Hume: nothing comes out of nothing, hence there is no event that is ultimately random. Even if our man was not the cause of the fire, the fire still had to have a cause (the lightning) which means that it was not random. This objection is fair enough, yet it is at the same time too general and too specific to be useful in the problem of legal causation. In first place, it is too general in the sense that it is concerned with the philosophical question of whether there is a general principle of causation: it answers in the negative, thus arguing that there is no ultimately random event. This may or may not be true, but it is certainly not the question that the legal system seeks to answer when pondering the problem of causation. Moreover, the objection is also too specific in the sense that it stops short at the first level of legal causation. It argues that the fire was not random, for it had to be caused by something. But then again, since the fire was indeed occasioned by a conjunction of several conditions, which include oxygen, flammable materials, the lightning, and so on, arguing that the fire was not random because it was occasioned by the presence of all these different conditions is not useful for juridical purposes. Legal causation requires the normative assessment of choosing among these conditions, singling out a causal link, and arguing that it is because of the presence of such cause that the event is not random.
John Stuart Mill faced the same problem, and took a sensible way out: given that all these conditions are necessary for the event to happen (oxygen, wood, and so on), then we simply cannot pick one as the cause of the event. The cause is for Mill the sum of all conditions; therefore ‘we have philosophically speaking no right to give the name of cause to one [condition] exclusively over the others’.Footnote 70 However, he further argued, not all choices are unsound. For this purpose, Mill distinguishes the ‘scientific’ from the ‘common notion’ of cause, the latter of which is useful ‘for the purpose of our present discourse’.Footnote 71 According to the first one, as we have seen, cause is the joint occurrence of all conditions. According to the second, however, a cause can be chosen among several conditions, a choice that is essentially dependent on the context and the purpose of the particular causal statement.Footnote 72 It is, as it were, a political moment of causation.
The common-sense notion of causation is the cornerstone of the legal approach to randomness. If the existence of causation signals the non-randomness of an event, then the criteria for choosing a causal link are also the criteria for determining randomness. Such criteria, however, are extra-legal: they depend on the context of who is making the choice.Footnote 73 By the same token, the place of chance in legal reasoning is also a function of extra-legal variables. If one of the conditions is chosen as the basis for the causal link, the choice will be correct, for the criteria for choosing one of those conditions are not given by legal language. The only thing that language does require is that, once the choice is made and the causal link is established, all events not related to it are to be deemed as random events, products of mere chance. Law has no way of providing a distinctively legal explanation as to why a certain condition should be chosen as a cause. Consequently it has no explanation either as to why an event should be considered as lacking a cause or, in other words, as merely random. Legal language requires that we look beyond the law for an answer on what is a cause and, thus, what is a random event.
3. Randomness and causation in international law
International law is no exception. This section will consider the notion of causation in international law and its implications for the role of randomness in international legal reasoning. The discipline lacks a distinctively ‘international’ notion of legal causation, and draws upon the preceding discussion when faced with the problem of cause. Consequently, international law features the same limitations as its domestic counterparts when faced with randomness and chance. It is not possible, this section will conclude, to assess random events using international legal language. As a result, international law's role in the risk society is undermined: instead of being an independent normative force, it is called on to become a mere broker of expert knowledge, as will be explained in the next (and final) part of this paper.
The issue of causation has been mostly ignored by international legal scholarship, and the discipline, to quote one of the few authors who have tackled the matter, ‘is the poorer for it’.Footnote 74 The lack of an expressly international theory of legal causation implies that the problem needs to be tackled, as it were, from the bottom up. We need to look at the substantive areas of international law where the problem of causation is more likely to rise, and see how the issue has been dealt with. The argument of this section is, essentially, that the problem has not been treated in a specifically ‘international’ fashion: international legal causation reproduces the two-level structure described in the previous section, wherein the ultimate choice of a cause is defined under a non-legal criterion.
3.1. Causation in the law of state responsibility
Perhaps the area that provides the most valuable insight is state responsibility. Even though causation and responsibility are commonly linked, they are two different matters.Footnote 75 Causation normally leads to responsibility – although it may not do so if, for example, harm was caused in self-defence; similarly, responsibility is normally based upon a positive assessment of causality – but it may not, as is the case with state responsibility. Under Articles 1 and 2 of the International Law Commission ArticlesFootnote 76 (ILC Articles), state responsibility is composed of two elements: (i) the attribution of an action or omission to a state under international law; and (ii) the breach of an international obligation of the state. These two elements constitute the internationally wrongful act which, under Article 1, ‘entails the international responsibility of the State’. For the purposes of our discussion the regime is notable in that it excludes any reference to causation of harm as a requirement of state responsibility.Footnote 77 Such exclusion was by no means coincidental. From 1924 the approach to state responsibility within the League of Nations was focused on the responsibility of states for damage to foreigners done in their territories.Footnote 78 In this framework, the presence of harm to aliens was required for the assessment of responsibility, thus making harm and its causation a fundamental element of the overall structure. This approach was adopted in 1958 by the first International Law Commission (ILC) Special Rapporteur on State Responsibility, Francisco García-Amador. Indeed, for García-Amador, the issue of state responsibility was so vast that the ILC should focus first on the issue ‘most ripe for codification’, namely the ‘responsibility of the State for injuries caused in its territory to the person or property of aliens’.Footnote 79 From then on, his work was focused on that issue,Footnote 80 as also were Sohn's draft articles, the other important landmark of that moment in the history of state responsibility. Indeed, causation as an element of responsibility was present in Article 1(1) of the 1961 Harvard Draft Convention on the International Responsibility of States for Injuries to Aliens, which stated,
A State is internationally responsible for an act or omission which, under international law, is wrongful, is attributable to that State, and causes injury to an alien.Footnote 81
This approach changed with the second ILC Special Rapporteur, Roberto Ago. By 1961 both the UN General Assembly and the ILC seemed convinced that the codification should not be limited to state responsibility derived from harm to aliens.Footnote 82 The expansion of the scope had important effects on the issue of causation. In his Second Report, Ago noted that ‘in addition to the two elements . . . that have been shown to be constituent elements of an internationally wrongful act which is per se a source of responsibility, reference is sometimes made to a third element, which is usually termed “damage”’.Footnote 83 However, to the Rapporteur's mind, the issue of damage was a matter of primary obligations derived from state duties towards individuals at the municipal level. These duties were unrelated to the rules of state responsibility which were, by definition, secondary rules, unconcerned with the actual (primary) source of responsibility.Footnote 84 Ago seemed to link damage intimately with the problem of causation. Thus when the first was dropped, the second obviously followed. Causation was excluded from the elements of state responsibility as, according to Ago, it was ‘inappropriate to take this element of damage into consideration in defining the conditions for the existence of an internationally wrongful act’.Footnote 85 Causation was never to return to the debate as an element of state responsibility. As finally adopted, the ILC Articles do not require harm to constitute responsibility except, of course, that the primary obligation requires harm for its breach, in which case causation would be an element of the international obligation that is breached.Footnote 86 Instead of being based on causation, state responsibility is based upon the notion of attribution.Footnote 87 This is not to say that state responsibility is a form of vicarious liability, but rather that it is wholly exhausted through the discussion of whether an internationally unlawful action or omission can be attributed to a state.Footnote 88 Once the attribution problem is solved, responsibility can be attributed, for causation has no bearing in this discussion.Footnote 89
Despite its exclusion from the structure of state responsibility, causation is not absent from the regime. It appears commonly in the framework of reparation, only to be swiftly discarded without much analysis. Consider, for example, the Tehran Hostages case, where the issue was for the most part ignored by the International Court of Justice (ICJ).Footnote 90 In this case, the ICJ decided to divide the hostage crisis into two distinct phases: the first one covering the overrunning in 1979 of the US embassy in Tehran, the taking of the hostages, and the inaction of the Iranian government, and the second comprising the facts after the completion of the occupation of the embassy and the seizure of the consulates in Tabriz and Shiraz.Footnote 91 According to the Court, in the first phase students acted independently and their action was not attributable to the Iranian government.Footnote 92 However, Iran's lack of action was instrumental for the success of the students' independent action.Footnote 93 Therefore both the militia's action and the government's omission were necessary, but not sufficient, conditions for the hostage crisis. The problem of causation was, thus, at the centre of the first phase of the Tehran Hostages case. However, the ICJ did not approach the matter from this perspective. It understood that this was not an issue of multiple conditions for harm, but argued instead that, regardless of causation of harm, Iran was responsible, for it had failed to protect the diplomatic compound in violation of its obligations under the 1961 and 1963 Vienna Conventions on Diplomatic and Consular Relations.Footnote 94
One of the few comprehensive approaches to causation in state responsibility can be found under Article 31 of the ILC Articles, which provides that the responsible state is obliged to make reparation for the injury caused by the internationally wrongful act. In this context, international legal causation becomes a familiar mix of ‘natural causation’ and ‘legal assessments’, which reproduces the two-level test of the domestic approach to legal causation, described above. Such strong reliance on domestic law was perceived by the ILC's Fifth Special Rapporteur, James Crawford, who pointed out in his Third Report that ‘in international as in national law, the question of remoteness of damage is not a part of the law which can be satisfactorily solved by search for a single verbal formula’.Footnote 95 The problem, to Crawford's mind, was that ‘“factual causality” is a necessary but not a sufficient condition for reparation. There is a further element, associated with the exclusion of harm that is too “remote” or “consequential” to be the subject of reparation.’Footnote 96 Crawford's argument overlaps here with our discussion on randomness: the exclusion of remote damage is, in effect, the selection of a cause among conditions in order to establish a causal link between conduct and harm, which is an all too familiar process at this point. How does one exclude remote harm under the law of state responsibility? There is no answer to that question: just as in domestic law, such assessment depends on the context. The Fifth Special Rapporteur correctly points that fact out:
[I]n some cases, the criterion of ‘directness’ may be used, in others ‘proximity’ or ‘foreseeability’. But other factors may also enter into the calculation: for example, whether the harm caused was within the ambit of the rule which was breached, having regard to the purpose of that rule.Footnote 97
Crawford's list is necessarily incomplete, for the process relies on non-legal variables that cannot be established in legal terms: they depend on the context of the specific case being decided, its characteristics and parties. The law of state responsibility shows that there is no distinctively international notion of legal causation: international law draws heavily from domestic legal systems on this point, and also borrows its inability to provide a distinctively legal explanation as to why a certain condition should be chosen as a cause.
Perhaps the most vivid example of this situation is one of the discipline's early landmarks, the Alabama Claims Arbitration case (1872).Footnote 98 The arbitration concerned a conflict between the United States and Britain on the occasion of the Civil War. Britain had pledged neutrality in the confrontation between Confederates and the Union; however, the Confederates managed to contract several ships with British boatyards. Of these Confederate ships, four inflicted particularly heavy losses to the Federal merchant fleet: the CSS Florida, the CSS Georgia, the CSS Shenandoah, and, of course, the CSS Alabama.Footnote 99 Once the Civil War ended with the victory of the Union, the United States demanded compensation from Britain, as it argued that the latter had violated its duties of neutrality by building the ships. The matter was to be settled through international arbitration based in Geneva, under the provisions of the 1871 Treaty of Washington.
In its complaint, the United States claimed an enormous amount of money as compensation, equivalent to six times the total of British annual expenditure at the time.Footnote 100 The problem was, essentially, indirect claims: the United States argued that it should be compensated for the prolongation of the war and, more importantly, for the loss of maritime commerce derived from the actions of British-built ships. Such claims were quite controversial: in his memoir, the Secretary the Senior American Counsel recounts how, in Britain, indirect claims attracted widespread attention and outrage,Footnote 101 while, even in the United States, the claims were deemed inappropriate: in an 1869 note, the editor of the American Law Review wrote that, although the British had certainly inflicted a deep loss upon the newly formed Union,
if we were to bring forward this great national loss as a matter of pecuniary claim, we should certainly find ourselves embarrassed with certain well-established, and not wholly pedantic, rules, familiar to the courts of law, as to remote and proximate causes of damage.Footnote 102
The indirect claims problem (alternatively called ‘national claims’, as the damage was allegedly suffered by the nation and not by individual American merchants) was no trivial matter: according to most commentators, there was a very palpable fear that, were the British to repudiate the Geneva Tribunal, war would ensue.Footnote 103 Facing this situation, the parties reached a political agreement between 15 and 18 June 1872, whereby indirect claims would be excluded from the litigation.Footnote 104 The Tribunal, in turn, issued an anomalous pre-award statement one day later, where it indicated its belief that
[indirect] claims do not constitute, upon the principles of international law applicable to such cases, good foundation for an award of compensation or computation of damages between nations, and should, upon such principles, be wholly excluded from the consideration of the Tribunal in making its award, even if there were no disagreement between the two Governments as to the competency of the Tribunal to decide thereon.Footnote 105
Thus was the problem of causation addressed in the Alabama Claims Arbitration. Aware of the delicate political context, the Tribunal excluded indirect claims through a void reference to principles of international law, holding in effect that national harm was an event of mere chance with regard to British actions. The Tribunal performed a clearly political selection among the conditions that were necessary for the harm, and found that there was no causal link between such harm and the assessed conduct. The arbiters are, however, not to blame: their choice among conditions was political because the legal notion of causation gave them leeway to choose; it simply did not provide tools to assess such causal link. As a result, the Tribunal found a satisfactory way out for both parties: although it denied indirect claims, it did award the United States US$15.5 million in gold in direct claims, which is a staggering award at any rate. It was around 5 per cent of UK annual expenditure at the time and, in terms of today's budget, would be equivalent to £150 billionFootnote 106 or the annual GDP of Sweden.Footnote 107 Certainly, the award was enough for both parties to declare victory and avoid war.Footnote 108
3.2. Causation in international trade law
International law's reliance on non-legal variables to frame causation and randomness can also be observed in international trade law, as revealed by the law and practice of the World Trade Organization (WTO). Three areas under WTO competence featuring clear connections to causal analysis are countervailing duties, anti-dumping duties, and safeguards. Despite substantial differences in their origins and effects, these three measures feature one single approach to causation: in order lawfully to adopt one such measure, a state must prove a causal link between subsidy, dumping, or increased imports, and an injury to the affected sector.Footnote 109 Such causal link is established through a two-step test, which is common to the assessment of all three measures:
1. the state must provide evidence of a link between the assessed conduct and the injury or threat thereof; and
2. it must rule out that the injury is not linked to other causal factors.Footnote 110
Two things in the test are worth noting here. The first is its treatment of multiple causations, which is explicitly dealt with in the second step. Unlike the law of state responsibility, international trade law openly recognizes the challenges of multiple causes, and imposes upon states the duty to disentangle them in order to impose countervailing or anti-dumping duties and to adopt safeguards. The problem is that, although the requirement is in the law and has been enforced,Footnote 111 there is no clear legal criterion for perform such disentanglement. That brings us to the second noteworthy aspect of the test: despite its wording, the test reveals the same understanding of legal causation featured in domestic systems and in the law of state responsibility. As in those cases, the assessment of causation in trade law is left to general statements of principle, without any substantive guidelines to deal with the matter.
This is easiest to illustrate by reference to the US – Wheat Gluten Safeguard case,Footnote 112 which features one of the most comprehensive approaches to causation found in WTO law. The case emerged in 1998, when the United States adopted definitive safeguard action on certain imports of wheat gluten, in the form of quantitative restrictions.Footnote 113 The measure was selective: while products from Canada (a party to the North American Free Trade Agreement (NAFTA)) were excluded, the quota did affect the European Communities (EC), whose members were the main exporters of gluten to the US market at the time.Footnote 114 Consequently the EC decided to take action against the United States, arguing that the latter had acted inconsistently with Articles I and XIX of the General Agreement on Tariffs and Trade (GATT) 1994Footnote 115 and with Articles 2(1), 4, 5, 8, and 12 of the Agreement on Safeguards (AS).Footnote 116 Claims under Articles 2(1) and 4(2) of the AS reveal the limitations of causation in international trade law. Regarding these claims, the Panel held that the United States had indeed violated such provisions by adopting the safeguards, as ‘the causation analysis applied by the US International Trade Commission (USITC) did not ensure that injury caused by other factors was not attributed to import’.Footnote 117
The United States attacked this finding in its appeal, arguing that ‘the meaning of the word “cause”, [in the relevant provisions] is “to bring about a result, whether alone or in combination with other factors – not “to cause on its own”’.Footnote 118 For the United States, ‘the plain meaning of “causal link” in Article 4.2(b) of the AS, first sentence, is consistent with this understanding of “to cause”’,Footnote 119 which, it argued, was satisfied by the USITC's investigation. By putting forward this argument, the United States brought before the WTO Appeals Body (AB) Hume's essential problem of causation when it argued that harm to its wheat gluten market was the result of multiple conditions, one of which was the increased quantity of imports. Given that Article 4(2) does not impose a requirement of ‘sole causation’, the USITC's choice of increased imports as a cause among several conditions was perfectly legitimate, for the harm was indeed ‘caused’ by it, regardless of whether it was able to ‘cause it on its own’. The conditions for such choice, however, remained indeterminate: from the American point of view, Article 4(2) does not require the isolation of imports for the establishment of a causal link, which, in any event, would involve ‘subjective speculation’.Footnote 120 On the contrary, the United States argued that Article 4(2)(b) of the AS requires that, once the causal link has been established, the competent authorities examine other causes of injury, in order to ensure that their effects do not sever the causal link between increased imports and serious injury.Footnote 121
The question before the AB was, thus, whether the Panel had erred in interpreting Article 4(2)(b) to mean that increased imports, in and of themselves, must be capable of causing serious injury.Footnote 122 If the Panel was right and Article 4(2)(b) establishes a sole-causation test, then there is ultimately no choice among conditions: only one condition can be chosen as the cause, and all other answers are simply wrong. If, on the contrary, the appeal's construction of the article was correct, then its test of causation leaves considerable leeway to establish the causal link, without providing any significant guidelines to that effect.
The AB agreed with the appellant, finding that Article 4(2)(b) does not require increased imports to be the sole cause of harm, and reversed the Panel's interpretation in this point.Footnote 123 For the AB, the requirement is satisfied if increased imports are among the conditions that led to harm, and are then singled out by the relevant authority as the cause of the injury, even if such increased imports did not, in and of themselves, cause the former.Footnote 124 In this sense the AB adopted an approach similar to Hume's ‘common sense’ causation (also present in domestic law, and in the law of state responsibility), whereby all conditions have to be present for the effect to take place, but one of such conditions may be legitimately chosen as the cause. Following the AB's reasoning, there is indeed choice among conditions, and the USITC was entitled to make its own in the wheat gluten case in order to fulfil the causation requirement under Article 4(2)(b). However, arguing that any choice made by the USITC would have been acceptable may be pushing the point too far; consequently, once the Panel's interpretation was reversed, the AB tried to assess the USITC's decision on substantive grounds.Footnote 125 Its effort to do so is also quite revealing of the limitations of causation in international trade law. The AB held that the USITC's decision was not in accordance with AS requirements, for it had failed adequately to evaluate the complexities of this issue and, in particular, the issue as to whether the increases in average capacity of wheat gluten production in the United States during the investigative period were causing injury to the domestic industry at the same time as increased imports.Footnote 126 This is a somehow problematic conclusion, for it seems to have required the USITC to prove the lack of a causal link between increased imports and injury, by proving the impact of augmented production capacity. That is clearly not what Article 4(2)(b) is about. And yet our discussion of causation helps to understand the AB's dilemma: on the one hand, since a mechanical formula of causation between increased imports and injury is not workable, the AB needed to leave space for choice among conditions by domestic authorities – it had to leave some room for them to interpret the causation requirement as they saw fit; however, on the other, the AB needed to impose some limitations to such interpretations, otherwise it risked turning irrelevant the whole construction of the causation requirement.
In the specific context of US – Wheat Gluten Safeguard, the AB was able to find a way out by effectively reversing the burden of proof on the issue of causation. Beyond the questions that this move may give rise to in WTO law,Footnote 127 it does show that international trade law is as ill-equipped as any other legal language to frame the issue of causation and chance. The requirement of causation under Article 4(2)(b) of the AS (and the equivalent provisions regarding anti-dumping and countervailing duties) is ultimately a question of whether the injury is a mere random event with regard to the increase of imports. International trade law seems unable to give a distinctively legal answer to that question. It seems to impose upon itself the mission of disentangling the problem of multi-causation, only to find itself forced to rely on non-legal variables to solve that very problem.
3.3. The precautionary principle
The elephant in the room in this assessment of causation in international law is the precautionary principle (hereinafter PP). This last subsection will argue that the principle is not useful in assessing the notion of causation in international law, because it is an exception to the notion. Now, as is well known, the principle (originally formulated in the early 1970s as Vorsogerprinzip in German environmental policy) seeks to define acceptable regulatory responses to uncertain and/or unknown harm. No single authoritative definition is available under international law; however, for the purposes of the discussion here, I shall follow the formulation contained in Principle 15 of the Rio Declaration on Environment and Development (1992),Footnote 128 according to which,
In order to protect the environment, the precautionary approach shall be widely applied by States according to their capabilities. Where there are threats of serious or irreversible damage, lack of full scientific certainty shall not be used as a reason for postponing cost-effective measures to prevent environmental degradation.
The Rio Declaration reflects the most basic form of the principle, according to which scientific uncertainty does not justify regulatory inaction.Footnote 129 At least two other versions exist (one holding that uncertainty justifies action, a second shifting the burden of proof to the proponent of risky activitiesFootnote 130), and Per Sandin has even proposed 19 different versions.Footnote 131 Common to all of them, however, is their treatment of causation – or rather the lack of it. The very essence of the principle is that rational choices can and should be made on the basis of uncertain relations of causation.Footnote 132 PP emerged as a reaction to perceived limitations of the traditional burden of proof as applied to cases dealing with environmental issues, specifically where the requirement of scientific certainty regarding the causal link between activity and harm could lead to unjustifiable regulatory inaction – inaction with potentially irreversible consequences.Footnote 133 The principle is thus not an application of the ‘but for’ notion of legal causation. It is an exception to such a notion. Ultimately, PP states that in certain cases it is perfectly acceptable to overlook causation: that is its added value. It implies that the legal notion of causation is too limited a notion to deal reasonably with certain risks, a fact that did not pass unnoticed by Beck. Here is Beck commenting on the requirement of strict proof of causation:
This is a good example of how ‘rationality’ can become ‘irrationality’, according to whether the same thought and action are seen through the frame of reference of wealth or risk production. The insistence on strict proof of causality is a central element of scientific rationality . . . At the same time, though, these principles stem from other contexts and perhaps even from a different intellectual epoch. In any case, they are basically inadequate for modernization risks . . . Anyone who insists on strict causality denies the reality of connections that exist nonetheless.Footnote 134
If the precautionary principle is an exception to the ‘but for’ legal notion of causation, then it is safe to conclude that it is not useful as a reflection of the notion of causation in international law. Beck's argument is, in essence, that what lawyers call the ‘precautionary principle’ is the appropriate starting point of a regulatory strategy in the risk society, and this in exclusion to strict causal analysis, which is ‘basically inadequate’ for the said purposes. PP is a different kind of answer to the challenges posed by risk societies, one that bypasses causation and seeks proxies, in order to allow rational decisions without relying on causal analysis. It is a parallel problem to causation, which will be dealt with when the general role of international law in risk societies is assessed.Footnote 135 For the time being, and as the examples of state responsibility and international trade law illustrate, international legal causation seems to be as doughnut-shaped as its domestic counterpart, a fact that undermines its relevance as an independent normative force in the context of the risk society, as we now turn to see.
4. The risk society and international law
The foregoing discussion featured two arguments that serve as the basis for drawing some conclusions on the role of international law in the risk society. The first argument holds that the fundamental characteristic of the risk society is randomness. The second proposes that, due to its understanding of causation, international law is effectively unable to deal with the notion of randomness. The conclusion to be drawn should not come as a surprise: international law is thus unable to tackle the challenges presented by the global risk society. Indeed, in a risk society international law faces randomness and uncertainty which need to be dealt with. For this purpose it deploys a set of argumentative structures that are intended to make sense of such uncertainty by recasting manufactured risks as external risks, thus making the former measurable in some way. Such structures are mostly based on causation, and reflect that notion's inner limitations: they are context-dependent, and are thus unable to provide a distinctively legal standard to evaluate the realities of risk societies. This section seeks to examine how, precisely, this inability expresses itself, and which place it seems to reserve for international law in risk societies.
To this effect, it is useful to consider the WTO's Agreement on the Application of Sanitary and Phytosanitary Measures (SPS Agreement).Footnote 136 As is known, the SPS Agreement seeks to strike a balance between free trade and the legitimate interest of states to adopt sanitary or phytosanitary measures – that is, measures to protect humans, animals, and plants against the risks of disease, pests, additives, toxins, and so on.Footnote 137 The SPS regime is of interest here because, in addition to standard non-discrimination and necessity conditions, it requires SPS measures to be based on (i) scientific evidence (Art. 2(2)), and (ii) risk assessment (Art. 5(1)). According to the AB, these two obligations are different, the first being broader than the second;Footnote 138 however, to date, most SPS disputes have been concerned with the second obligation, for SPS measures try to prevent harm that is not immediately inflicted on, nor perceived by, human beings – a typical risk assessment matter. As a consequence, the SPS regime has become a paradigmatic battleground of the risk society, where conflicting views of risk rationale often collide.Footnote 139 For all its complexity, the SPS Agreement is relatively straightforward in its legal structure. Member states can lawfully adopt SPS measures through three methods: (i) choosing measures that conform to international standards (Art. 3(2) of the SPS Agreement); (ii) adopting national measures based on international guidelines or recommendations (the so-called ‘harmonized standards’) (Art. 3(1)); and (iii) adopting domestic measures that impose a higher lever of protection than international standards, if there is a ‘scientific justification’ for doing so (Art. 3(3)). The adoption of SPS measures under the third option is at the origin of most litigation. To date, five cases have been decided by the Dispute Settlement Body concerning SPS measures, and the literature on the issue is enormous.Footnote 140 Such wide debate will prove, in turn, useful for testing the argument on the role of international law advanced in this paper.
Under Article 5 of the SPS Agreement, SPS measures should be based on risk assessment, which in turn has to be based on scientific principles and sufficient scientific evidence.Footnote 141 This could be seen as a clear confirmation of Beck's risk theory; however, one needs to be wary of drawing simple conclusions from the presence of the word ‘risk’ in this context. Despite first impressions, the fact that the SPS regime refers to ‘risk’ does not necessarily imply that adjudication on SPS measures shares any of the concerns related to the risk society. On the contrary, most SPS litigation seems to be focused on assessing measures from the perspective of external risk: future events that can be quantitatively evaluated, through a scientific method, in order to reduce the probability of harm. The AB was clear in this regard when it held in Australia – Salmon that ‘the ‘risk evaluated in a risk assessment must be ascertainable risk; theoretical uncertainty is not the kind which, under Article 5.1 is to be assessed’.Footnote 142 For Beck, the restrictive approach taken by legal language is not only insufficient to assess the challenges of the risk society, but is also a way of actually increasing risk, as ‘insisting on the purity of scientific analysis leads to pollution and contamination . . . what results then is a covert coalition between strict scientific practice and the threats to life encouraged or tolerated by it’.Footnote 143
There seem to be two positions in conflict: on the one hand, the SPS regime, requiring strong scientific evidence to justify SPS measures and, on the other hand, Beck, arguing that such a requirement is inappropriate, as it actually increases risk. This is, however, a false dichotomy.Footnote 144 Beck's approach is more pragmatic that it sounds: while he argues that notions of causation and risk underlying international legal language are inappropriate to address manufactured risks of pollution, he also argues that the same notions are appropriate to address manufactured risks of global terrorism.Footnote 145 How is it possible to make sense of this apparent contradiction? Our previous discussion of causation may shed some light on the matter. Legal reasoning, we have seen, is unable to provide a distinctively legal answer to the problem of causation and chance. The assessment of causation is context-dependent, and calling international law to perform such an assessment is but an appeal to perform an evaluation of context. Therefore the assessment of terrorism risk is necessarily different from the evaluation of the risk of pollution, as each risk's context is necessarily different from the other's. Beck's choice is wise in that sense: the role of international law in the risk society is marked by its theory of causation, but such a theory is actually context-dependent, therefore the role of international law in the risk society is, as well, context-dependent.
In turn, the AB seems to follow a similarly sensible path. For all its emphasis on strict scientific evidence, the AB seems surprisingly flexible in its approach to risk assessment and the interpretation of Article 5.2 of the SPS Agreement. Such a flexible approach can be gleaned from the EC – Hormones case.Footnote 146 As is well known, the case relates to a complaint brought by the United States and Canada against an EC ban on imports of meat and meat products from cattle treated with certain hormones for growth-promotion purposes.Footnote 147 In its report, the Panel held a strict view on the matter (reflecting the approach so strongly attacked by Beck) and argued that ‘an assessment of risk is, at least for risks to human life or health, a scientific examination of data and factual studies; it is not a policy exercise involving social value judgments made by political bodies’.Footnote 148 In this way, the Panel drew a clear line between risk assessment and risk management, the former being a strictly scientific endeavour, the latter being a choice of policy.Footnote 149 The AB, however, reversed the Panel on that point, famously stating,
[T]here is nothing to indicate that the listing of factors that may be taken into account in a risk assessment of Article 5.2 was intended to be a closed list. It is essential to bear in mind that the risk that is to be evaluated in a risk assessment under Article 5.1 is not only risk ascertainable in a science laboratory operating under strictly controlled conditions, but also risk in human societies as they actually exist, in other words, the actual potential for adverse effects on human health in the real world where people live and work and die.Footnote 150
True enough, the AB has doubted the current legal status of the precautionary principle,Footnote 151 but its approach seems to be less mechanical than is normally understood. Despite its strong statements of principle regarding scientific evidence, the AB seems to accept a more nuanced process of risk assessment by accepting that context is relevant for the test. This approach drew heavy fire, especially because of its consideration of allegedly irrational and unscientific variables in the process, such as consumer anxiety and fears.Footnote 152 Others, on the contrary, argued that the AB's approach was too mechanical and failed to consider other non-statistical variables.Footnote 153 The truth is perhaps in the middle, very much in consonance with Beck's own pragmatic approach. Considering the Panel's restrictive position with regard to Articles 5(1) and 5(2), the AB was once again faced with a dilemma: it could certainly not assert the ‘irrationality’ of consumer fears and other non-quantitative variables, but neither could it simply ignore them, for they were important aspects of the EC's argument and political context. The AB then took the only sensible way out: it admitted the importance of context in risk assessment, but held that, in any case, the EC had not fulfilled such a requirement. The EC had had only presented evidence regarding risks in cases where hormones had been administered under good veterinary practice, but had not presented evidence concerning risks where they had not been administered in accordance such practice.Footnote 154 Given the failure to produce such evidence, the EC's risk assessment was deemed insufficient, making its measures a violation of Articles 5(1) and 5(2).
The SPS debate is useful for illustrating that, despite their apparent differences, both Beck's and the AB's reading of risk assessment seem to share the belief that contextual analysis is fundamental for a viable risk assessment. This reliance is also expressed in our previous discussion of causation in international law. In a risk society, international law faces randomness and uncertainty which need to be dealt with, as a non liquet due to lack of evidence is in effect not an option. International law then deploys a set of argumentative devices, which are intended to make sense of uncertainty by recasting manufactured risks as external risks, thus making them somehow measurable and knowable. These devices are mostly based on the notion of causation, and reflect in turn its inner limitations. Take, for example, Article 5 of the SPS Agreement. Through the requirement of risk assessment and scientific justification, Article 5 aspires to translate manufactured risk (characterized by uncertainty) into a probabilistic problem of external risk. This translation, however, is not mechanical, as it is ultimately based on the notion of causation. Therefore the AB has to refer to variables that are external to the SPS Agreement, in order to interpret specific measures in their context, and determine their compliance. In this sense, although expressed in legal terms, the assessment is not distinctively legal as it necessarily relies on non-legal variables to solve the particular case. Both Beck and the AB seem to agree on this point. However, as we shall soon see, by consistently relying upon non-legal variables to face uncertainty, international law declines its role as an independently normative criterion for evaluating the risk society.
Now this is not a matter of legal hermeneutics. Although the inherent indeterminacy of language may play a part in international law's inability to provide answers in the risk society, the ultimate problem here is not that expressions like ‘sufficient scientific evidence’ are semantically ambiguous.Footnote 155 Even if there were no linguistic indeterminacy at all, the AB would still have to refer to non-legal variables to assess whether a measure complies with the SPS Agreement. Neither is this a problem of politics ‘contaminating’ legal reasoning. The unregulated area of causation is not a failure of the law, nor a violation of the law: it is the law. International law, as a legal language, is structurally unable to deal with the notion of uncertainty. Therefore, regardless of the interpretative style we adopt to deal with randomness and risk, international law will still fall short of framing the issue. SPS jurisprudence provides a good example of the irrelevance of interpretation for the matter of uncertainty. In EC – Biotech Footnote 156 the Panel seemed to give a more textual interpretation of Articles 5(1) and 5(2) of the SPS Agreement, thus tightening up the test featured in EC – Hormones. Indeed, besides analysing the EC's moratorium on biotech products,Footnote 157 the Panel also addressed safeguard measures enacted by Austria, France, Germany, Greece, Italy, and Luxembourg against such products.Footnote 158 To do so, it applied a pretty stringent test to risk assessment procedures, finding fault in all those that did not match precisely the definition contained in Annex A(4) of the Agreement.Footnote 159 Here is the Panel addressing one of Austria's safeguards:
We need not determine whether relevant scientific evidence was or is insufficient for Austria, and if so, whether this would be a relevant circumstance. Even if this were the case, the flexibility which the phrase ‘as appropriate to the circumstances’ may in some situations provide does not relieve Austria from the requirement in Article 5.1 to base its safeguard measure on a risk assessment which meets the definition of Annex A(4).Footnote 160
This is clearly a world away from EC – Hormones. The interpretative technique has changed, and the Panel's rhetoric is also substantially different.Footnote 161 Despite these differences, the approach to uncertainty remains unchanged from one decision to the other. Both decisions are faced with a problem of uncertain harm, and each decision appeals to a different source for certainty: context in EC – Hormones and scientific reasoning in EC – Biotech. However, although at opposite ends of the spectrum in their hermeneutic approach to a semantically ambiguous provision, both decisions are equivalent in one sense: neither of them actually expects certainty to come from the adjudicating instance itself.
Ulrich Beck argues that it is precisely because such limitations exist in causal analysis that we need the precautionary principle.Footnote 162 Indeed, as Ewald has pointed out, precaution ‘rounds out the agenda of the “risk society” in several ways’.Footnote 163 Yet PP is not more conducive than causation to reaffirming the role of international law as an independently normative force, and this is mainly due to two reasons. To begin with, Beck's alternative is to accept statistical correlation as a proxy of causation,Footnote 164 a proposal that ultimately keeps causal analysis in the background and simply shifts the burden of proof from the plaintiff to the defendant. If statistical correlation exists, it falls upon the defendant to prove that there is no causal link between her activities and correlated harm. To be sure, one could always call for strict liability and provide that, facing correlation, defendants shall never have the chance to rebut causation.Footnote 165 There is, however, little reason to believe that PP necessarily implies strict liability in environmental cases, for, even in its most extreme form, the principle would hardly require action in the face of uncertainty. As a result, Beck's proposal either keeps the ball in the field of causation, or implies a regulatory strategy that seems all but unworkable in practice – thus undermining even further international law's claim for independent normative pull.
A similar problem can be gleaned from François Ewald's account of the principle in risk societies. Ewald's notion of professional risk builds upon the divorce of causation from attribution.Footnote 166 An appropriate reaction to risk, then, is to allow attribution without causation, a reading that ultimately leads Ewald to rely on examples from the Code Civil that are, essentially, solutions of strict liability in the face of risk.Footnote 167 However, precaution is to Ewald much more than a mere system of strict liability. To his mind, uncertainty renews the privileges of the politician as decision-maker: if strict cost–benefit analysis does not provide the answer, then ‘the decision still belongs to the politician rather than to the expert, and is the result more of an ethic, of the respect of certain procedures, than of a morality linked to the application of an existing framework’.Footnote 168 Two things should be noted in Ewald's approach. The first is that his view of PP implies an important margin of appreciation for the decision-maker, which simply replicates the area of uncertainty existing in legal causation. Thus although causation and PP are two different (and parallel) answers to the challenges of the risk society, both of them are ultimately dependent on context for their application. The fact that it is international law that is calling for such dependency is evidence of its capitulation as an independent normative force in a risk society, an aspect that will discussed in the last section of this article.Footnote 169 The second interesting aspect of Ewald's approach to PP is his reliance on rules of procedure. Ewald seems to believe that, even in the context of complete uncertainty, rational answers may be reached if precaution is joined by appropriate decision-making procedures. Such rules of procedure may constitute mechanisms of democratization in regulatory practices, or may be heuristics for Ewald's re-empowered politician; however, they are not a mechanism for reinforcing international law's independent normative pull. Ewald's account argues the importance of following procedural rules, but fails to provide an explanation as to why those rules should be followed in the first place. Admittedly, that is not Ewald's task, yet international lawyers are bound to take note that such an approach simply presumes the pull towards compliance with procedural rules.Footnote 170
There is one final reason why PP is not conducive to reaffirming the role of international law as an independently normative force. The principle is yet another factor of uncertainty in a risk society. The choice to regulate pre-emptively due to precaution is, in itself, a source of new and unforeseeable risk. If uncertainty is the defining trait of the risk society, then regulation of uncertain risk is perhaps as risky as regulatory inaction.Footnote 171 In its extreme version, PP is a ‘shoot first, ask questions later’ policy, with all the hazards that such a policy may entail. Does a precautionary ban on a given activity entail higher risks than the lack thereof? Would a precautionary ban on GMOs be a manufactured risk in itself? If we are to take the risk society seriously, the only coherent answer is that we cannot know; there are no default answers. Thus the precautionary principle becomes an argumentative tool that necessarily requires contextual analysis for its implementation. PP is no one-size-fits-all solution to environmental problems. Just like causation, it is a generally phrased criterion whose application to particular problems is dependent on the given context. The precautionary principle, too, features a hole in the doughnut. The issue is not that PP leads in the wrong direction, but rather that it leads in no direction at all.Footnote 172 It, too, gives up its claim to certainty deriving from legal language itself.
4.1. Brokering knowledge, outsourcing authority
In a global society where uncertainty is the rule, international law works as a hinge between different sources of certainty. It serves as a broker of knowledge, calling upon relevant experts to provide the necessary authoritative explanation that will, in turn, infuse certainty into legal analysis. This move has been subject to increased scrutiny, as the interaction between law and science grows closer.Footnote 173 Some commentators see in this trend a threat to good regulatory practices,Footnote 174 and some others fear that it may restrict the possibility of member states adopting the appropriate level of protection for their populations.Footnote 175 Beyond these valid concerns, it seems important to examine the consequences of this trend for international law itself. If international law is indeed a mere broker of knowledge in risk societies, how is the ideal of an international rule of law affected? Such is the question that this (last) section seeks to consider. To do so, it seems convenient to start where we left off. Due to its notion of causation, international law seems unable to deal with randomness and chance. That is, however, precisely what is required from it by risk societies. Hence, considering that it cannot simply shy away from the challenge, international legal language refers to non-legal variables in order to provide concrete answers in specific cases. By doing so, international law seems to work as a broker of knowledge, calling upon the expertise that is needed in a particular case.
Interaction between international law and authoritative expertise is controversial due to its perceived lack of transparency. Ulrich Beck noted early on what he perceived to be the ‘technocratic challenge to democracy’,Footnote 176 and submitted:
This is what will decide the future of democracy: are we dependent on the experts for everyday detail in issues concerning survival, or does the culturally manufactured perceptibility of hazards restore us to the competence to judge for ourselves?Footnote 177
Thus presented, Beck's position comes close to what Sunstein has called ‘risk populism’ – that is, the view that ‘tends to distrust experts and to think that in a democracy, government should follow the will of the citizenry rather than that of a technocratic elite’.Footnote 178 Such a view contrasts with so-called ‘technocratic’ views, which believe that ‘ordinary people are frequently ill informed and urge that the task of regulators is to follow science and evidence, not popular opinion’.Footnote 179 This discussion, however, is not relevant for assessing the role of international law in a risk society. Indeed, two different axes of discussion should be discerned here: one, the ‘populism v. technocracy’ dilemma, examining the role of technical expertise on (democratic) governance, and a second, that examines the impact of such expertise on the development of an international rule of law. Although obviously connected, the two axes are different in an aspect of crucial importance for our discussion here: while there is a stark contrast between populists and technocrats with regard to governance, both stances share the belief that regulation's normative authority derives from a source that is external to international legal language (democratic legitimacy, scientific rationality, and so on), thus confirming the idea that international law is a mere broker of knowledge, lacking any normative pull in and of itself. Therefore, even though debate over ‘technocracy’ is bound to have important implications for other areas of international law (e.g. the reform of decision-making proceduresFootnote 180), such relevance is lost for present purposes, as, ultimately, both populists and technocrats see international law as a hinge between different sources of certainty.
Now, how is the brokering of knowledge performed? Drawing from Kennedy's eloquent image, the interaction takes place in the background of global politics, where ‘other people than those who seem to be in charge are making the real decisions’.Footnote 181 These ‘other’ people can be appropriately described as an ‘epistemic community’,Footnote 182 that is, a ‘network of knowledge-based experts or groups with an authoritative claim to policy-relevant knowledge within the domain of their expertise’.Footnote 183 In the risk society, epistemic communities perform a double role vis-à-vis legal reasoning, following closely the two central traits of manufactured risk. Indeed, as we saw above, manufactured risks are such because they are created by human hand (and are thus not ‘external’). Moreover, risks are also ‘manufactured’ because they only exist through human knowledge – they are, we said, ontologically subjective. Expertise taps into this dual nature in its relation with international law. First, following risk's subjective nature, epistemic communities establish risks that are to be tackled by legal reasoning. It is only through their knowledge that manufactured risks come into being, and become relevant for adjudication. We have discussed earlier the example of cigarettes. Smoking is by nature a hazard, but it only becomes a risk through the intervention of the medical community's expertise. In this specific sense, epistemic communities create manufactured risks – without them, only unconnected occurrences are to be observed. Once created, manufactured risks need to be processed through international legal language. This second interaction deploys epistemic communities' second role with regard to legal language. According to Haas, the epistemic community's role is, essentially, to frame issues for policymakers: they reveal hidden cause–effect relations in complex issues and help policymakers define their self-interest.Footnote 184 International adjudication relates to expertise in a similar way. Under uncertainty and perplexed by technical complexity, international lawyers turn to the epistemic community, which gives certainty by explaining what causes generate which effects. Faced with deep uncertainty, and structurally unable to deal with it, international legal reasoning turns to different networks of experts, borrowing from their authority to increase certainty, dealing thus with randomness in a risk society.
It should be noted, however, that international lawyers do not form, in themselves, an epistemic community. For one thing, international law as a field lacks, in Haas's words, ‘the social authority or legitimacy of the technical authority commanded by epistemic communities’;Footnote 185 moreover, international lawyers seem to perceive their craft as an instrument for policymaking. Thus, while epistemic communities frame issues and influence policymakers, legal experts justify policy choices.Footnote 186 The relation between international law and epistemic communities is a one-way street, where international law draws from other areas of expertise, in order to provide the answers that the risk society demands from it. By doing so, international law is effectively abandoning any aspiration to act as an independent normative force within global society. It seems constantly to require assurance from other disciplines in order to be sure that it is doing the right thing. Such is the consequence that its notion of legal causation has for international law. International legal reasoning is dependent upon non-legal variables to meet the challenges of randomness, which is the defining trait of risk societies. As a result, international law's independent normative pull seems negligible: its worth in the risk society seems not self-given, but recognized by others.
This situation is not due to Beck's second modernization. This article does not intend to prove that the risk society contributes to international law's inability to deal with randomness. Such inability existed before the second modernization, and certainly would have continued, regardless of the events accounted for by Beck and Giddens. However, those events did present new challenges to the old concept of causation, thus shining the spotlight on the latter's limitations. Thus what was a theoretical delicacy only relevant for tort lawyers suddenly became the central aspect of a veritable political battle, as is evidenced by the example of SPS regulation. This change of emphasis did come with the second modernization, for it exponentially increased the number and visibility of all those events that legal language was not (and had never been) ready to face. The matter is, therefore, one of emphasis rather than one of substance: the law's limitations remained unchanged, but the risk society brought with it challenges that could not be dealt with. These challenges became visible, in turn making the law's inherent limitations more visible, thus making such issues politically relevant.
To be sure, international law is not alone in this tight spot. Domestic law, too, is unable to frame randomness, as its notion of causation is similarly dependent on non-legal variables; it is also of limited use when faced with the challenges of the risk society. However, unlike its international counterpart, its independent normative pull is not undermined by this limitation.Footnote 187 To understand such asymmetry, one has to consider that, as we have discussed, risk societies seem to find law useful only if it achieves a certain policy goal.Footnote 188 The risk society features, if you will, an ‘instrumentalist’ view of the law.Footnote 189 In domestic law this view is balanced by a strong system of hierarchy in legal sources and a centralized enforcing system, which gives a certain value to the norm in the abstract, if only because it has fulfilled certain procedures of norm creation. Thus the independent normative pull of domestic law is unaffected by its lack of inability to deal with risk and randomness, even if confronted by the risk society's instrumentalist view. International law, on the contrary, has nothing with which to balance such a view. It lacks a hierarchical theory of sources, and it lacks a central system of enforcement. Thus, when confronted with the risk society's instrumental view (and being as unable as domestic law to deal with risk and randomness), international law's very claim to independent authority comes into question. In this context, even if both systems feature the same limitation, domestic law's worst-case scenario is to be perceived as non-responsive to a challenge. For international law, the stakes are much higher: in a risk society, international regulation is authoritative only if supported by the relevant epistemic community – hence, its worst-case scenario is to be perceived as utterly irrelevant and be consequently put aside as a regulatory entity. In a risk society, limitations of legal causation affect the quality of domestic law, but may ultimately undermine the very relevance of the international legal system.
In a risk society international law seems to lack an inherent normative value. Such a view gives up its faith in formal rules for the purpose of achieving a greater good, but it then fails to provide a criterion for assessing that objective. And, although such an approach can be praised for its flexibility and pragmatism, it does seem to place an awful lot of trust in policy goals that may, but may not, be ultimately acceptable.Footnote 190 Thus if we want the freedom to ban GMOs, we argue that a strictly textual reading of the SPS Agreement is inappropriate, for we have a greater good to achieve and cannot get entangled in mere formalities; but at the same time we trust an anti-terrorist convention to make a difference by giving a ‘clear definition of terrorism’, and there, we argue, international law is indeed an invaluable tool.Footnote 191 By the same token we try to create a biotech market by praising the SPS Agreement in its literal complexity, but also doubt whether international law is useful for assessing the risk of global terrorism.Footnote 192 In the risk society, international law seems to be a book that is taken off the shelf when a good quote is needed.
4.2. Pondering consequences
A conclusion to be drawn from reasoning is that international law is not the most appropriate response to the challenges of the risk society. Perhaps Beck's call for more international regulation is simply the wrong prescription for the right diagnosis – after all, there is no evidence as to why we should even bother with considering the role of international law to begin with. Maybe the risk society is a paradigm change that goes beyond international legal language and will eventually make it obsolete. We would then be better off leaving the problem to epidemiologists, economists, and intelligence experts. ‘Your argument is convincing’, one would hear, ‘so please be silent. This is no lawyer's problem.’
This article makes no normative argument in favour of international law's inherently superior position to deal with the risk society. On the contrary, it is an effort to prove that international law is having a hard time tackling such challenges and that perhaps other languages are more suited to react. The problem is that such an alternative is ultimately moot. Although current regulatory mechanisms seem to be put into question by the risk society, the idea of regulation itself is not. Answers are developed through new regulatory mechanisms, which in any case require an independent normative force for their very existence. As a consequence, voids and contradictions in legal language end up, for better or for worse, at the centre of any discussion on ‘new regulation’. A good example of this dynamics is environmental regulation. As new challenges emerged from pollution and the growing evidence of its consequences, it became clear that so-called ‘command and control’ regulation was becoming ineffective.Footnote 193 Thus in a policy wave that overlapped with risk society scholarship, new regulatory approaches were developed. The new harvest included economic instruments such as emissions trading, Pigouvian taxes,Footnote 194 and even ‘information-based’ instruments such as public disclosures of pollution discharges, or Ecolabels.Footnote 195 All these regulatory mechanisms go beyond mere legal enforcement, and yet they still depend on legal language for their existence. A policy of emissions trading is certainly not a mere ‘command and control’ problem, but its design and implementation do bring to the game all the limitations of international legal language. And this is so because, when push comes to shove, policymakers are hard-pressed to come up with answers to the risk society that bypass international law for good. Therefore when this article tries to make sense of international law's limitations, it is not calling for its dismissal, but rather recognizing its pivotal role and trying to open up spaces in order to improve its performance as an independent normative force, and not a mere broker of knowledge. For keeping that latter role, although well deserved due to its limited notion of causation, is likely to have important effects on the possibility of achieving any significant version of an international rule of law.
5. Conclusion
Causation can be the central aspect of a legal outcome (as in anti-dumping duties), or it can be only marginally important (as in state responsibility). However, when the context of adjudication is marked by uncertainty and randomness, legal causation becomes a fundamental element, not only for the specific legal outcome, but for the general perception of the role of law in that context. Our global risk society is indeed marked (even cursed) by unsolvable uncertainty. Causation thus becomes a fundamental element in understanding the role reserved to international law in that society. We have seen, however, that such a role is modest at best, as the law requires external reassurance to confirm its normative worth: it seems to renounce its value as an independent criterion of validation, a fact that may have an impact on the possibility of advancing towards an international rule of law. Indeed, international law as an independent normative force seems to be withering away. And this, I fear, is no random outcome for, as Borges has pointed out, ‘every humiliation is an act of penitence, every failure a mysterious victory, every death a suicide’.Footnote 196
