The Yale Law Journal


Opting out of the Law of War: Comments on Withdrawing from International Custom

08 Dec 2010

Professors Curtis Bradley and Mitu Gulati have written a rich and interesting paper with a bold conclusion supported by historical and normative arguments. I find myself unpersuaded by either set of arguments. Most of my comments concern their reading of the historical sources, which they use to show both that the Default View was indeed the traditional doctrine of customary international law (CIL) and that the Mandatory View carries a disagreeable colonialist legacy. I criticize the former argument in Part I and the latter in Part II. But I also reject their claim that the Mandatory View makes CIL too “sticky” and hard to change. Taking the law of war as the principal example, I suggest in Part III that stickiness may be a virtue, because the alternative to supposedly outmoded customary law is likely to be no law rather than better law.

The law of war is the elephant in the room for any proposal arguing that states should be able to opt out of customary international law at will. It receives only passing and cursory mention by Bradley and Gulati, but it has been the most controversial part of customary international law at least since 9/11, as the United States locked horns with allies, NGOs, and the International Committee of the Red Cross (ICRC) over the interpretation of international humanitarian law in the “war on terrorism.”

A recent example illustrates how contentious the role of the customary international law of war has become in the United States. In Al-Bihani v. Obama, a panel of the District of Columbia Court of Appeals asserted boldly that it is “mistaken” to suppose that the President’s war powers “are limited by the international laws of war.” A sign of how explosive this view is may be seen from the fact that the government itself disavowed it in its response brief to a petition for en banc review. Ultimately, seven of nine judges backed away from the assertion, finding it unnecessary to dispose of the case. The seven judges did not elaborate—their denial of the petition for en banc rehearing was only one sentence long—but three concurring opinions did. Judge Janice Rogers Brown had written the original panel opinion, and she vigorously reaffirms it. In Judge Brown’s view, the majority’s reticence to say plainly where it stands on the domestic force of the international law of war is symptomatic of a far larger issue. The majority, Judge Brown suspects, shares an “intuition” with certain scholars of an internationalist bent. It is the “intuition that domestic statutes do not stand on their own authority, but rather rest against the backdrop of international norms.” Judge Brown rejects this intuition, and suspects that the majority does not. She insists that her panel’s holding still has “binding authority.”

Judge Kavanaugh weighs in with a voluminous concurring opinion defending the Al-Bihani I conclusion of the domestic inefficacy of international law of war and, indeed, of customary international law in general. He defends a thesis originally proposed by Curtis A. Bradley and Jack Goldsmith, that after Erie v. Tompkins customary international law is no longer part of U.S. law. This is a thesis that the Supreme Court rebuffed in Sosa v. Alvarez-Machain; Judge Kavanaugh, however, reads Sosa as an endorsement rather than a rebuff of the Bradley-Goldsmith thesis.

Judge Brown may well be right that the majority wants to reject these radically anti-international-law propositions. Perhaps it ducked the issue because its members could not agree on an alternative account of the role of the customary international law of war in constraining the President. The majority opinion is conspicuously minimal, and even though it refuses to endorse the panel’s attempt to bury the customary law of war, the opinion certainly says nothing to praise it.

Without discussing the merits of any of these debates, I note only that they are symptomatic of profound U.S. disquiet with the customary law of war. Those who oppose giving the law of war a place in U.S. domestic law marshal their arguments with considerable passion. I agree with Judge Brown that the real bone of contention is this larger issue. My point is that—given how fraught the issue has become in U.S. discourse—the customary law of war should be at the top of the agenda in considering Bradley and Gulati’s proposal. I fear that the Default View has the potential for weakening the already fragile protections in the law of war—not only in legal doctrine but on the battlefield.

That is a good reason for carefully considering whether Bradley and Gulati get the history right. Did the Default View actually represent the mainstream of international law until recently? Does the Mandatory View actually have an obnoxious colonialist pedigree? As I shall show, Bradley and Gulati’s principal historical source for the Default View, Emer de Vattel, never accepted that states can opt out of the law of war. It is instructive to observe how carefully Vattel restricted the Default View to exclude the law of war, largely because these laws are based on natural law. Grounding the law of war in natural law is Vattel’s way of arguing that they ought to be sticky.

I. The Historical Sources of the Default View

Let’s start with the history—both the writings of the publicists and the nineteenth-century proclamations of the Supreme Court. Bradley and Gulati quote several striking passages that on their face represent explicit statements of the Default View. Read in context, though, their significance is less clear, because the background assumptions of these writers—about both the scope and the force of customary international law—differ substantially from present-day law. Some of these authorities maintain that states may opt out of customary law only because in their view custom does not actually bind states in the first place. Today that assumption is almost universally rejected, including by Bradley and Gulati, who make it clear that they are not attacking the very idea of binding customary international law. But without the assumption, this argumentative pathway to the Default View closes, and so these authorities have little contemporary relevance. Vattel, on the other hand, advocates the Default View with no such background assumption. An important virtue of Bradley and Gulati’s article is that it calls attention to these neglected dicta of Vattel. But, on closer inspection, Vattel restricts the Default View’s scope to a subclass of what today is thought of as customary international law, excluding some of the most important present-day rules (which Vattel emphatically did not regard as optional). Notably, these include the law of war.

A. Vattel and Customs that Track Necessary Law

Because Vattel is the classical theorist to whom Bradley and Gulati devote the most attention, it is worth going more deeply into Vattel’s views. We must begin with his basic taxonomy of international law. Like many classical writers, Vattel distinguishes natural law from positive law, and he rests the latter entirely on state consent. The positive law of nations includes conventional (treaty) and customary law, but Vattel adds a third and less familiar category: the so-called “voluntary” international law. Voluntary international law consists of rules flowing as corollaries from the requirement that states respect each other’s sovereign liberty and equality—a requirement that forms “the foundations of their natural society.” For Vattel, these three categories of the positive law of nations (voluntary, conventional, and customary) “all proceed from the will of nations,—the voluntary from their presumed consent, the conventional from an express consent, and the customary from tacit consent.” Unlike express and tacit consent, the presumed consent on which voluntary international law rests is a legal fiction that imparts consensualist credentials to the principles of sovereign equality and liberty, even without evidence that states have tendered it.

The term “presumed consent” might suggest that state consent to voluntary law is a presumption that states can overcome, for example by announcing that they do not accept a rule of voluntary law. However, that is not Vattel’s view. On the contrary, as Bradley and Gulati note, Vattel does not permit states to opt out of the voluntary law. That is because the voluntary law:

  consists in rules of conduct and of external right, to which nations are, by the law of nature, bound to consent; so that we are authorised to presume their consent, without seeking for a record of it in the annals of the world; because, even if they had not given it, the law of nature supplies their omission, and gives it for them. In this particular, nations have not the option of giving or with-holding their consent at pleasure: the refusal to give it would be an infringement of the common rights of nations.  

In sum, voluntary law is positive law; it derives its character as positive law from the presumed consent of states; but presumed consent is itself a requirement of natural law. Rules of voluntary law are, we might say, an indirect derivation from natural law.

In addition to these three categories of positive international law, significant rules we today include within customary international law are said by Vattel to be “necessary” law, a category that he derives directly from natural law. For example, Vattel characterizes the rule of foreign ministerial immunity as necessary law. In general, “nations are absolutely bound to observe [the necessary law of nations],” and “nations [cannot] reciprocally release each other from the observance of it.” Thus, like voluntary law, necessary law follows the Mandatory View, not the Default View.

As Bradley and Gulati note, however, Vattel also asserts that necessary law is not in and of itself positive law, and the obligation it imposes is moral rather than legal. But what if a rule of necessary law has also been “positivized” through custom, so that the custom is supported not only by the tacit consent of the parties but also by their moral obligation under necessary law? Although Vattel says nothing explicit about such cases, a careful reading reveals that he does not support the Default View for rules of customary international law that implement necessary law. Notably, in the key passages where Vattel asserts the Default View, he is careful to stipulate that a custom to which it applies is “indifferent in itself.” In another passage he refers to custom that is “in its own nature indifferent” or at most “useful and reasonable.” He clearly does not regard necessary law as indifferent in itself (quite the contrary). Regarding foreign ministerial immunity from prosecution in the courts of another state, which today we consider customary international law, Vattel not only asserts that the rule is necessary law, he dismisses the significance of a state opt-out from the correlative customary rule (the “established practice,” in Vattel’s phrase):

  If it be true that there was a king of Spain, who, from a desire of arrogating to himself a jurisdiction over the foreign ministers resident at his court, wrote to all the christian princes, informing them that if his embassadors should commit any crime in the places of their respective residence, it was his pleasure that they should forfeit all of their privileges, and be tried according to the laws of the country—one solitary instance is of no weight in an affair of this nature; nor have his successors on the Spanish throne adopted a similar mode of thinking.  

Here we have a perfect example of a rule of customary international law that is also part of the necessary law, and we see that Vattel does not analyze it under the Default View.

Next consider the law of war. Vattel argues that all acts of violence in unjust wars are forbidden to the unjust side, a prohibition he describes as the necessary law of nations, or of the law of nature, which nations are strictly bound to observe: it is the inviolable rule that each ought conscientiously to follow.” This establishes a moral, if not legal, obligation. Furthermore, at most one side in a war can be the just side. But, in an interesting twist, Vattel next argues that the voluntary law of war (which, recall, has positive-law status in his system) deviates from the necessary law of war in both these respects: the former but not the latter holds that regular war . . . is to be accounted just on both sides” and “whatever is permitted to the one in virtue of the state of war, is also permitted to the other.” It follows that the jus in bello rules are also voluntary law because “[t]he whole is to be deduced from one single principle,—from the object of a just war.” And indeed, Vattel explicitly confirms that jus in bello rules, like the prohibition on hostile acts undertaken without military necessity, belong to the voluntary international law.

Vattel’s point is that even though one state may in fact be the unjust aggressor and its target may be the just defender, it is too dangerous and too contrary to the equal sovereign liberty of states to permit one state to assert that it has justice on its side and the other does not. Therefore, international law creates the legal fiction that both states are waging just war, even though under natural law at most one state can have justice on its side. The fiction that all states have equal claims to justice is precisely the voluntary law of nations. It is still reflected today in the immunity of states from adjudication in another state’s courts, under the principle par in parem non habet imperium (equals have no dominion over equals).

Recall that Vattel’s view was that states cannot opt out of the voluntary law. Thus the most important parts of today’s customary law of war would in Vattel’s eyes be subject to the Mandatory View, not the Default View.

So for Vattel, the category of customary international law for which he articulates the Default View is significantly narrower than contemporary customary international law because it excludes voluntary rules like the law of war and even “established practices” that represent necessary law, like foreign ministerial immunity. The only rules of customary international law to which Vattel applies the Default View are those that are “indifferent in themselves”—customs where the need for some default rule is evident but it does not much matter which of several alternatives custom settles on. Vattel does not support Bradley and Gulati’s case for the historical primacy of the Default View very strongly. States cannot opt out of some of the most important rules of customary international law, and in the case of those rules Vattel would more accurately be described as an opponent of the Default View. Notably, as I have argued, Vattel denied that states can opt out of the law of war.

Let us now more briefly examine some of the other authorities Bradley and Gulati cite to support their claim that the Default View represents the historical mainstream before the twentieth century.

B. Burlamaqui and the Primacy of Natural Law

Burlamaqui, who like Vattel holds a natural law view, presents a different case because he denies that customary international law is obligatory at all: “these customs, are neither of themselves, nor universally, nor always obligatory . . . . We cannot therefore avoid appealing always to the law of nature, the only one, that is really universal, whenever we want to judge whether the customs established between nations have any obligatory effect.” Burlamaqui seems to hold the Default View only in the trivial sense that if customary international law is not obligatory in the first place, states can opt out of it. Burlamaqui is not a CIL supporter who accepts the Default View; rather, he is a CIL skeptic.

C. The U.S. Supreme Court on Customary International Law

For similar reasons, I am not persuaded that the U.S. Supreme Court decisions Bradley and Gulati cite support their case. Brown v. United States, for example, does so only in the sense that it takes the Burlamaquian position that customs are not binding laws at all:

  [U]sage is a guide which the sovereign follows or abandons at his will. The rule, like other precepts of morality, of humanity, and even of wisdom, is addressed to the judgment of the sovereign; and although it cannot be disregarded by him without obloquy, yet it may be disregarded. . . . It is not an immutable rule of law, but depends on political considerations which may continually vary.  

And so, states may opt out of “usages” for the theoretically trivial reason that they are never genuinely obligated by them in the first place. This background assumption makes Brown of doubtful relevance to a theory of customary law as binding law.

The same is true of Justice Chase’s opinion in Ware v. Hylton, which asserts that custom “is only obligatory on those nations, who have adopted it.” This is a weaker idea of customary international law than the traditional view—which is also the prevailing view today—that customs can bind even states that have not adopted them. Neither Brown’s nor Ware’s arguments can be read as an assertion that a state may opt out of a binding legal obligation.

For a different reason, Schooner Exchange v. McFaddon represents a weaker principle than the Default View. It deals only with a special case of customary international law, namely a customary rule that “[i]n almost every instance” is stipulated by treaty. From that fact, Justice Marshall infers that in the unusual case of a state that has not entered any such treaty, sovereign “assent [is] necessarily implied.” He continues, “Without doubt, the sovereign of the place is capable of destroying this implication,” namely the implication that even without the treaty the sovereign has tacitly consented to the customary rule. This argument would not apply to customary international law in general, as Bradley and Gulati seem to conclude. Rather, it applies only to implicit agreements where explicit treaties have established the presumptive rule. For, clearly, the “implication” Justice Marshall refers to is the implication that the state has tacitly consented to the same rule to which other states have explicitly consented through treaties. Thus Justice Marshall is not saying that as a blanket matter states can opt out of customary rules. He is saying only that states can indicate that they never opted in.

Finally, Bradley and Gulati’s revisionist reading of The Paquete Habana as an assertion of the Default View is unsustainable. The Paquete Habana finds that “[i]nternational law is part of our law,” but courts should apply “the customs and usages of civilized nations” only “where there is no treaty, and no controlling executive or legislative act or judicial decision” to the contrary. Bradley and Gulati propose to read this qualifying language as a version of the Default View. On their proposed reading, “any of the three branches of the federal government would have the authority to make the opt-out decision”—which is, indeed, the only version of the Default View that the phrase “executive or legislative act or judicial decision” could be said to articulate.

But is it conceivable that The Paquete Habana was asserting that the judiciary has unilateral authority to opt the United States out of a rule of international law? That would make The Paquete Habana one of most important cases in U.S. history on the separation of powers (and inconsistent with all subsequent cases asserting that the foreign affairs power rests with the political branches). It is one thing for a court to state, “We find that this rule does not bind the United States,” but quite another to state, “We find that up to now this rule bound the United States, but declare that henceforth it does not.” Obviously, courts have the authority to find that domestic law precludes judicial enforcement of international law—indeed, that is the standard way of reading the qualifying language in The Paquete Habana. But the proposition that courts have the authority to pull the United States out of a rule of international law on their own would be a momentous and very aggressive assertion of judicial authority over foreign affairs, on a par with a court claiming the power to withdraw the United States from a treaty. That the Supreme Court established such authority in The Paquete Habana in a brief and cryptic dictum that no subsequent court ever read this way is quite implausible. Ergo, the qualifying language Bradley and Gulati quote from The Paquete Habana is not a version of the Default View. Furthermore, even though The Paquete Habana cites Brown, it explicitly rejects the Burlamaquian view in Brown that customary international law is not binding law: by asserting that international law is part of our law, the Court can mean nothing other than that it is legally binding.

II. Was the Mandatory View a Pro-Colonialist Innovation?

There is another piece to Bradley and Gulati’s historical story—more speculative, but one that is central to their normative concerns. This is the hypothesis that the Mandatory View displaced the Default View for colonialist or imperialist reasons. As evidence, they point to the 1905 edition of Oppenheim’s influential treatise, which articulates the Mandatory View. They suggest that Oppenheim was worried that new, “uncivilized” states would opt out of customary international law and that he propounded the Mandatory View as a way to bind new states to the rules of their predecessors.

This explanation strikes me as too anachronistic to be plausible. It might be right had Oppenheim articulated the Mandatory View four decades later, when decolonization was on the horizon, but it is hard to believe that new postcolonial states were on his mind in 1905. What new states? It seems vanishingly unlikely that Oppenheim foresaw 1950s African decolonization or Indian independence at the zenith of European imperialism. If the Default View was indeed the accepted doctrine of CIL, it is hard to believe that Oppenheim would alter it simply to ensure that hypothetical states whose existence Europeans did not imagine would abide by European customary law.

Professor Kelly, who Bradley and Gulati cite in support of their conjecture, refers to “new, non-Western nations such as Turkey and Japan” and “states that emerged from the collapse of the Ottoman empire.” Presumably, though, Kelly is referring to the Turkish republic and Middle Eastern states created post-World War I; he could hardly have meant Ottoman Turkey, which was not a new nation and which made treaties with Europe continuously from the fourteenth century through the Congresses of Paris and Berlin in the nineteenth. The Balkan states carved out of the Ottoman Empire by the Treaty of Berlin in 1878 (Serbia, Romania, and Montenegro) were Europe-oriented Christian states whose willingness to participate in international law was not in question. Of course, when Oppenheim wrote his treatise, the Ottoman Empire was palpably in collapse. But from the vantage point of 1905, Ottoman dissolution would lead to its provinces being absorbed by the hungry great powers, not by their becoming independent states. Even during World War I the great powers secretly agreed to carve up the eastern Ottoman Empire into sectors under their direct control and the control of puppet governments.

Japan is a more plausible example, although it too was not a new state. The Institut de Droit International conducted a study of customary international law in 1875, dealing with the question of whether European customary international law could function “in the Orient,” given the different beliefs and legal traditions; it concluded that for major Asian states, including Japan, the answer was yes. Perhaps Oppenheim disagreed, and propounded the Mandatory View to lock Asian states into European rules. But that is pure speculation. It seems more straightforward to take Oppenheim at face value: if he propounded the Mandatory View rather than the Default View, perhaps that is because the Default View was not as firmly established as Bradley and Gulati suppose.

Bradley and Gulati acknowledge that the shift to the Mandatory View might have come later, perhaps 1930 or even the end of World War II. If so, their explanation of the Mandatory View as a colonialist device might gain more traction. But the very fact that Oppenheim, the most influential treatise writer of the era, asserted the view decades earlier makes it hard to accept the late date for the shift.

III. Stickiness and the Law of War

Now let me turn to what I see as Bradley and Gulati’s principal normative argument for the Default View, which is based on considerations of institutional design. Bradley and Gulati argue that a signal disadvantage of the Mandatory View is that it might make rules of customary international law too sticky, so that they remain stuck in inefficiency. To turn to the example that I find most significant, they speculate that “[m]any of the existing rules governing war are likely outdated, in that they were designed for different types of armed conflict than the types of conflicts we see today.” Regrettably, they do not elaborate.

Which rules, specifically, are they talking about? Most of the laws of armed conflict are treaty-based rules that have become customary international law through the familiar dynamic of widespread ratification and acceptance into state practice (treaty ratification itself providing evidence of opinio juris). The most prominent examples of such rules are Geneva Convention provisions requiring decent treatment of captives, and the jus in bello provisions in Additional Protocol I of the Geneva Conventions, such as the requirements of discrimination and proportionality in collateral civilian damage set out in Articles 51 and 57. Recently, a number of commentators have suggested that these rules are outdated because they are not adapted to asymmetrical conflicts. Commentators have recommended that proportionality should be reconfigured, or that the category of enemy combatants should be expanded to include civilian supporters of terrorists several steps removed from participation in hostilities. And, famously, then-White House Counsel Alberto Gonzales told the President that under a “new paradigm” of war, “Geneva’s strict limitations on questioning of enemy prisoners” are “obsolete,” while other provisions are “quaint”; and the United States government rejected rules protecting “unlawful enemy combatants” from humiliating and degrading treatment or trials with inadequate procedural protections, until the Supreme Court’s 2006 Hamdan decision.

Bradley and Gulati seem to accept that some (unspecified) subset of these rules is outdated and inefficient and that the Mandatory View mistakenly prevents states from opting out of them. Yet there are a few reactionaries among us who think that this particular inefficiency is a good thing. We want the rules against mistreating enemy combatants or causing disproportionate injury to civilians to be sticky because we loathe detainee mistreatment and civilian casualties even more than we loathe inefficiency. We think that a more permissive rule, efficient or not, would be a moral setback; and a view of customary international law that makes protections of captives and civilians easier to eliminate because someone thinks they are inefficient is not one we embrace. It took millennia of bloodshed and butchery before nations settled on these prohibitions as a matter of custom; they represent a singular moral achievement. To allow states to opt out of them on grounds of their possible inefficiency—or inefficiency in one state’s eyes—seems cavalier.

Of course, the stickiness of these rules also inhibits the formation of customary rules that mandate better treatment of enemy combatants or civilians than current International Humanitarian Law (IHL). Perhaps this is what Bradley and Gulati have in mind. But states do not need to opt out of the current rules in order to move to a practice of treating civilians or enemy combatants better than the rules require. No legal prohibition prevents them from doing so now, and if they begin doing so they may form a new state practice without having to opt out of the old one. It seems far more likely that the result of opt-outs would be an unraveling of the fragile IHL regime and its replacement by a rule of “we’ll do what we want to do”—in other words, no rule at all.

A significant consequence of the rules of CIL is that states train their militaries in them. The United States, which did not ratify Additional Protocol I, nevertheless accepts portions of it as legitimate statements of customary international law and builds these rules into its law-of-war training and JAG practice. One virtue of a hierarchical, bureaucratized organization like an army—obviously, hierarchy and bureaucracy have drawbacks as well—is that once rules are accepted, they eventually turn into training manuals, protocols, rules of engagement, JAG advice to commanders, and directives at all levels. To take a small example, soldiers in the U.S. Army carry with them a plastic wallet card, titled “Call to Duty: Boots on the Ground,” that has the seven Army Values and four Principles of the Warrior’s Ethos printed on the front, and ten Soldier’s Rules printed on the back. The Soldier’s Rules are basic precepts of the law of armed conflict, and the wallet card is not a casual idea: it is the product of bureaucratic processes that developed all these doctrines (values, ethos, rules) and studied methods of disseminating them through the ranks.

Now suppose that the Default View came to be accepted in place of the Mandatory View of customary international law. And suppose that, faced with the threat of terrorism, the United States decides to opt out of, say, the proportionality rules. The United States is not a party to Additional Protocol I, so opting out of the customary proportionality rule would not leave a treaty rule as a safety net. Inevitably, military training and JAG practice would change to reflect the altered law. But what would it change to? Almost certainly, the opt-out would lead to a period of flux, with lawyers taking stabs at formulating new rules that conform to the latest operational doctrines. In other words, instead of the law of war acting as a constraint on military behavior, we would experience a period in which military behavior determines the U.S. law of armed conflict.

Furthermore, the United States is deeply involved in training other states’ militaries, and so it seems likely that some might follow the U.S. lead. I see the Default View moving not toward better customary rules, but toward no rules—figuratively (but maybe literally as well), a Period of Warring States replacing Law’s Empire before settling down into a new body of customary law, if it ever does. The predictable result, it seems to me, would not necessarily be a more efficient law of war; but it would almost certainly include more dead civilians.


Ultimately, the worry about the Default View is that it weakens rather than strengthens international law. And, ultimately, the intuition behind the Mandatory View is that a system of law in which you can divest yourself of legal obligations simply by announcing that henceforth you are no longer bound is not really a system of law at all. These are two sides of the same concern. Gerald Postema has described customs—not just customary international law, but customs more generally—as a commons, and I think that is right. Commons are maintained by participation, and participation carries normative force. Too many opt-outs can destroy the commons. Bradley and Gulati assure us that there would not be excessive opt-outs even if the Default View prevailed. But, as the law of war example suggests, in some cases even one opt-out by a very powerful state might be one too many.

David Luban is University Professor and Professor of Law and Philosophy at the Georgetown University Law Center. The author presented an earlier version of this response at the Washington University Workshop on Public International Law on February 5, 2010. He would like to thank Curtis Bradley, Mitu Gulati, and Carlos Vázquez for helpful discussion of the first draft.

Preferred citation: David Luban, Opting out of the Law of War: Comments on Withdrawing from International Custom, 120 Yale. L.J. Online 151 (2010),