Customary International Law Symposium: The Stubborn Tenacity of Secondary Rules

Customary International Law Symposium: The Stubborn Tenacity of Secondary Rules


Monica Hakimi’s new article, “Making Sense of Customary International Law,” is my favourite kind of scholarship: bold, critical, revisionist, tendentious. Too few scholars are brave enough to confront the sacred cows of public international law this forthrightly, and for that reason alone Hakimi deserves our thanks and praise.

That said, I disagree with nearly every word in Hakimi’s article.

An adequate response would require an article of its own. In what follows, I simply want to sketch two of the most important objections to her theory of customary international law (CIL): (1) her supposedly “sociological” perspective bears no resemblance to how states, international courts, and other important international actors actually view CIL; and (2) instead of banishing secondary rules from CIL, she simply replaces the “rulebook” theory’s admittedly fuzzy secondary rules with secondary rules that are even more fuzzy.

How Global Actors View Custom

The essence of Hakimi’s theory is the idea that CIL “does not consist of a set of secondary and primary rules” (1520). In other words, the classic “two-element” theory of custom formation is wrong – the idea, to quote the ILC’s Draft Conclusions on Identification of Customary International Law, that “[t]o determine the existence and content of a [primary] rule of customary international law, it is necessary to ascertain whether there is a general practice that is accepted as law.” “Put simply,” Hakimi insists, “the rulebook conception reflects what many people imagine CIL to be, but it does not describe what global actors use and receive as CIL in the everyday practice of law. It does not reflect what CIL ‘is’ as a real-world sociological phenomenon” (1491). On the contrary, “[g]lobal actors regularly engage in this practice – they invoke, argue about, and justify their decisions in CIL – even though CIL does not operate like a rulebook” (1536).

Hakimi calls this a “straightforward descriptive claim” (1491), but it is anything but – particularly with regard to the most important “global actor” within the international system, states. There is no question that states disagree, often intensely, about whether state practice and opinio juris are sufficient to establish a primary rule of CIL. There is also no question that states routinely take issue with how international courts, NGOs, and even other states understand and apply CIL’s secondary rules. But Hakimi does not provide even one example of a state rejecting the “rulebook” theory of CIL as a whole – i.e., describing a given norm as CIL even though it does not view that norm as a valid and binding primary rule produced by the operation of secondary rules. The closest she comes is in her discussion of whether the so-called “unwilling or unable” (U/U) test for self-defence against non-states actors qualifies as a primary rule of CIL, where she emphasizes that states have not always reacted to supposedly defensive uses of force in a manner that is consistent with the rulebook theory – i.e., by either affirming or rejecting their legality (1533). Those non-binary responses, however, do not indicate that the states in question reject the idea that CIL consists of valid and binding primary rules. They simply illustrate that states can support a particular defensive use of force for political reasons without simultaneously wanting that support to count towards U/U’s legality – or vice-versa.

Hakimi’s failure to cite an example of a state that rejects the rulebook theory is not surprising, because states uniformly embrace the idea that CIL consists of valid and binding primary rules produced by identifiable secondary rules – they simply disagree about how the secondary rules should be understood and applied. After all, as Hakimi herself acknowledges (1490), the General Assembly endorsed the ILC’s Draft Conclusions without a vote, brought them “to the attention of States and all who may be called upon to identify rules of customary international law,” and encouraged “their widest possible dissemination.” The two-element theory is at the heart of the Draft Conclusions. States also routinely rely on the rulebook theory when they argue about the customary status of particular rules. The U/U debate discussed above is one example. Another is the heated debate at the Sixth Committee over the ILC’s conclusion that government officials do not enjoy functional immunity for international crimes. Some states supported that conclusion; others forcefully rejected it. All, however, based their position on the rulebook theory of CIL.

Even the US, so often sceptical of CIL, firmly embraces the rulebook theory. Just consider its formal response to the ICRC’s study of customary IHL. After noting in the “Methodology” section that “[t]here is general agreement that customary international law develops from a general and consistent practice of States followed by them out of a sense of legal obligation” (444), the response then discussed four rules that the US believes do not qualify as custom because the ICRC overestimated their support in state practice and/or opinio juris. At no point did the US challenge the rule-based nature of customary IHL; it simply disagreed with the ICRC about which norms qualified as rules.

The rulebook theory, in short, does not simply reflect “what many people imagine CIL to be.” It reflects what states imagine it to be. Hakimi’s rejection of the rulebook theory thus implies that states are suffering en masse from some sort of CIL false consciousness: CIL “does not consist of a set of secondary and primary rules,” even though states uniformly believe otherwise. That is a strange claim given that states are the authors of international law.

Although Hakimi fails to provide examples of states rejecting the rulebook theory of CIL, she does spend considerable time trying to show that other actors in the international system do not accept the theory. One example she cites is the Bashir case at the ICC, in which the Appeals Chamber held that heads of state do not enjoy personal immunity from arrest by another state when the arresting state is executing an ICC arrest warrant. According to Hakimi, “[i]n making this decision, the Appeals Chamber did not apply the two-element test or any other secondary rule. It did its own thing” (1510). But that is contradicted by the very sentence from the appeals judgment that Hakimi quotes immediately before she makes her claim: “the onus is on those who claim that there is such immunity in relation to international courts to establish sufficient State practice and opinio juris.” That is a traditional application of the “two-element test,” however contestable the application itself might be. (In my view, the Appeals Chamber erred in deciding that “no immunity before international courts” was the customary baseline.) So the Appeals Chamber was not doing “its own thing.” It was doing precisely what the rulebook theory of CIL requires.

Hakimi’s understanding of the ICRC’s Interpretive Guidance on the Notion of Direct Participation in Hostilities is similarly problematic. She claims that “the ICRC did not contend that its positions satisfied the secondary rules of CIL or had the status of primary rules. It said only that they “reflect the ICRC’s institutional position[s] as to how existing IHL should be interpreted” (1513). The ICRC’s institutional position, however, is precisely that “only state agreements (treaties) or state practice followed out of a sense of legal obligation on a certain issue (custom) can produce binding law” (7). The rules set out in the Interpretive Guidance, therefore, reflect the ICRC’s understanding of what international law requires – as evidenced by, inter alia, its claim that “[u]nder customary and treaty law, civilians directly participating in hostilities, as well as persons assuming a continuous combat function for an organized armed group belonging to a party to the conflict, lose their entitlement to protection against direct attack” (69). This is not an example of a global actor rejecting the rulebook theory of CIL. It is an example of such an actor knowing that it has no authority to produce “a text of a legally binding nature” (7) and thus endeavouring to apply the rulebook theory as faithfully as possible.

These are not “gotcha” comments concerning the ICC and ICRC. The point is that all global actors – states, international courts, NGOs, and so on – believe that valid and binding primary rules of custom exist and that those rules are determined by the operation of identifiable (if difficult to apply) secondary rules. So it is not true that “the rulebook conception… does not describe what global actors use and receive as CIL in the everyday practice of law.” Global actors – particularly states – do not always speak the language of CIL. But when they do, they invariably treat CIL as a rulebook.

The Inevitability of Secondary Rules

Indeed, it would be surprising if it were otherwise. CIL is not simply a language that states can use to convince other states to accept their actions as lawful (though that is obviously one of its uses). More fundamentally, it provides a set of valid and binding primary rules that can be used to resolve legal disputes, particularly in judicial fora – international or national. If there were no primary or secondary rules of CIL that took the form of a rulebook, such dispute resolution would be impossible.

Consider a case before the ICJ. State A claims that the relevant customary rule is X. State B insists that the relevant customary is Y. How should the ICJ resolve that dispute? The traditional “rulebook” answer is that the Court must apply the secondary rules of CIL – the two-element theory – to determine which primary rule, X or Y, is valid and binding. That is what the ICJ has done in cases ranging from North Sea Continental Shelf to Nicaragua to Jurisdictional Immunities. The ICJ might get the rule wrong by misapplying the secondary rules. But unless it intends to throw up its hands, declare a non liquet, and send everybody home, it has to apply them.

Hakimi never makes clear what a court should do when faced with the need to resolve a dispute over a customary rule. At times, she seems to deny the very existence of rules of CIL that are capable of “authoritatively resolving what ought to be done” (1536) – which would mean that all legal disputes over custom must end in a non liquet. That skepticism is implied by Hakimi’s repeated insistence that CIL cannot be defined in “rule-like terms” (1505). And it is even more explicit in her discussion of the “unwilling or unable” test, where she says categorically that “[b]ecause multiple CIL positions are in play, no one of them operates as a rule” (1527).

Most of the time, though, Hakimi suggests that there are at least some norms of CIL sufficiently rule-like to resolve disputes that turn on their existence (or non-existence). Thus she says “in some circumstances, constraining official discretion through settled conduct rules is both possible and important” (1534). And elsewhere that “[s]ome CIL norms do operate like primary rules and can be said to satisfy a version of the rulebook’s secondary rules” (1520). And still elsewhere that “[s]ome of [CIL’s] conduct norms display rule-like levels of clarity and stability” (1491).

These statements, however, raise a critical question: how does Hakimi distinguish between norms of conduct that “operate like primary rules” and norms of conduct that don’t? The rulebook approach, of course, relies on the secondary rules to distinguish between primary rules of CIL and norms that are soft law: the primary rules are those norms – and only those norms – that are supported by sufficient state practice and opinio juris. Hakimi, however, specifically rejects the rulebook’s reliance on the secondary rules, insisting that even if “[s]ome of [CIL’s] conduct norms display rule-like levels of clarity and stability… it is not because they satisfy certain secondary rules” (1491). And why not? Because “CIL does not have secondary rules” (1506).

As it turns out, though, Hakimi underestimates secondary rules’ tenacity. Here is what she says about how primary-rule-like norms of conduct come about:

The lack of secondary rules in CIL does not mean that “anything goes.” It means that what goes is not determined by secondary rules. The status of a given normative position within CIL depends instead on how global actors interact with it over time. To what extent do these actors invoke, regard, and use the position as CIL, rather than ignore or challenge it? Thus, as a practical matter, those who want a position to have traction as CIL must find support for it. They must earn authority for the position from other participants in the CIL process. Insofar as global actors broadly accept and treat a position as CIL, it becomes entrenched. At some point, it might even garner enough support to operate like a conduct rule (1510).

This paragraph describes a process of “entrenchment” that is eerily similar to the two-element theory of custom. Under the rulebook theory, the status of a primary rule is determined precisely by how states “interact with” the normative position in question “over time.” Do states “use the position as CIL, rather than ignore or challenge it” – i.e., do they engage in a practice (for permissive rules) or abstain from engaging in a  practice (for prohibitive ones)? Do states “regard… the position as CIL” – i.e., do they view it as legally binding or as simply a matter of comity? Are states able to convince other states to “broadly accept and treat a position as CIL” – thereby establishing that the supporting state practice and opinio juris is sufficiently widespread and representative? If so – if “there is a general practice that is accepted as law,” in the words of the ILC – then a valid and binding primary rule of custom exists.

The difference between the rulebook theory of CIL and Hakimi’s theory, therefore, is not that the former relies on secondary rules while the latter does not. On the contrary, Hakimi’s theory is no less dependent on secondary rules – those rules are just fuzzier than the ones used by the rulebook theory and result not in valid and binding primary rules but in norms that “operate… like primary rules.” I have no idea what the difference is between a primary rule of CIL and a norm that operates “like a primary rule.” But I suspect it has no practical import.


Hakimi could have written an article that argued the ambiguity and indeterminacy of the “rulebook” theory’s secondary rules make it extremely difficult to determine valid and binding primary rules of CIL, thereby undermining the practical utility of the theory. I would have wholeheartedly endorsed that argument. Or she could have claimed that states argue about the legitimacy of particular actions in ways that do not always reduce to classic CIL categories. I would have agreed with that, too. Instead, she wrote an article claiming that “global actors” reject the rulebook theory and that CIL does not make use of secondary rules. The former claim is descriptively false and the latter is analytically unsound, as Hakimi’s own theory of custom, which is no less dependent on secondary rules than the rulebook theory, demonstrates.

This post has been quite critical, so let me end on a more enthusiastic note. I think Hakimi’s insistence that international lawyers pay closer attention to the “argumentative dimension” of CIL (1533) is incredibly important and generally underappreciated. As a discursive strategy, rote invocations of CIL – like rote invocations of international law generally – are of limited use: it will be the rare situation indeed where a claim of illegality will convince a state, particularly a powerful one, not to engage in act that it views as necessary to protect its interests. That does not mean CIL does not take the form of primary rules, nor does it mean that the validity of primary rules is not determined by CIL’s secondary rules. But it does mean that insofar as international contestation is not simply a war of all against all – which it too often is – it is far more important to convince “the people in power [to] justify how and toward what ends they use their power” (1533) than to convince them that they are acting illegally. Differently put, to paraphrase Hakimi (1527): binding or not, identifiable or not, the primary rules of CIL are not coming to save us.

Print Friendly, PDF & Email
Articles, Featured, General, Public International Law, Symposia
No Comments

Sorry, the comment form is closed at this time.