… that the appropriate reparation in this case consists in the obligation of the United States of America to provide, by means of its own choosing, review and reconsideration of the: convictions and sentences of the Mexican nationals …by taking account both of the violation of the rights set forth in Article 36 of the Convention and of paragraphs 138 to 141 of this Judgment.

Failure of the United States to provide such review and reconsideration, would be a breach of its international obligations. In the case of the LaGrands this had been irrevocably the case as they had both been executed in the United States by the time the judgment was delivered. Contrary to the situation in the LaGrand case, all of the individuals whose rights were at issue in the underlying Case of Avena and Certain Other Mexican Nationals were still alive by the time the Court had delivered its judgment on 31 March 2004. In fact, Mr. Medellin was the first for whom an execution date had been set since, which was to be carried out on August 5, 2008. It is in this context that one should put Mexico’s June 5, 2008 filing of a request for interpretation of the Avena judgment, and the request for an order of provisional measures accompanying it, which most importantly sought to halt Mr. Medellin’s execution. With only two months left before Mr. Medellin’s scheduled execution date, the ICJ managed to hold oral arguments and produce an order of provisional measures within the very short timeframe of 41 days.

Judging from Mexico’s initial request, many familiar with Article 60 of the Court’s Statute had early on counted Mexico out. After all, Mexico had the difficult task to convince the Court that it was not simply trying to have another round of proceedings on the United States’ failure to implement Avena. Rather it had to demonstrate that, in the words of Article 60, there was a “dispute as to the meaning or scope of the judgment” allowing the Court to “construe it upon the request of any party”.

Mexico’s initial request seemed ill-equipped to achieve that goal, asking the Court to declare that

..the obligation incumbent upon the United States under paragraph 153(9) of the
Avena Judgment constitutes an obligation of result as it is clearly stated in the
Judgment by the indication that the United States must provide “review and
reconsideration of the convictions and sentences” but leaving it the “means of
its own choosing;”
and that, pursuant to the foregoing obligation of result,
(1) the United States must take any and all steps necessary to provide the
reparation of review and reconsideration mandated by the Avena
Judgment; and
(2) the United States must take any and all steps necessary to ensure that
no Mexican national entitled to review and reconsideration under the
Avena Judgment is executed unless and until that review and
reconsideration is completed and it is determined that no prejudice
resulted from the violation.

Taking the request at face value, it seemed that, unless the United States would be gullible enough to maneuver itself into the position of arguing that the review and reconsideration mentioned in paragraph 153(9) of the Avena Judgment did not refer to an obligation of result (but one of conduct), there would be no way for Mexico to demonstrate the existence of a dispute.
Moreover, the two ancillary requests appeared rather redundant if compared to the Court’s findings in Avena, which I quoted above.
In fact, all the United States seemed to need to do was to find a way to gracefully admit that it would indeed be in breach if the state of Texas executed Mr. Medellin, while at the same time accepting that the obligation upon them was of course an obligation of result – case closed.

That is essentially the position the United States took during oral arguments. Of course, it also offered arguments seeking to avoid the order of provisional measures, and sought to showcase the efforts of the US government to implement Avena.
Mexico on the other hand elaborated its initial request, arguing that the conduct of the United States (for example the Federal Government’s lack of further initiatives to implement Avena and the state of Texas’ scheduling of Medellin’s execution) confirmed that the United States truly held the interpretation that paragraph 153(9) imposes only an obligation of means. Hence, in addition to its earlier request, Mexico asked in its submissions at the end of the oral proceedings, that the Court “specify that the obligation to take all steps necessary to ensure that the execution not go forward applies to all competent organs of the United States and all its constituent subdivisions, including all branches of government and any official, state or federal, exercising government authority”.

None of this to my mind seemed to overcome the general difficulty Mexico faced – there was simply no dispute. But to those who may have agreed with me, the Court’s Order of provisional measures provided a surprise. After a lengthy explanation of the meaning of “dispute” as used in different provisions of the Court’s statute, which demonstrated that a “contestation” in French can be broader than a “différend” (the latter being the French term used for example in Article 38(1) of the Statute), the Court sided with Mexico and ordered provisional measures, reasoning that

…whereas, while it seems both Parties regard paragraph 153 (9) of the Avena Judgment as an international obligation of result, the Parties nonetheless apparently hold different views as to the meaning and scope of that obligation of result, namely, whether that understanding is shared by all United States federal and state authorities and whether that obligation falls upon those authorities;
56. Whereas, in light of the positions taken by the Parties, there appears to be a difference of opinion between them as to the meaning and scope of the Court’s finding in paragraph 153 (9) of the operative part of the Judgment and thus recourse could be had to the Court under Article 60 of the Statute;

This finding can be criticized in several regards: on the one hand, the Court demonstrably found that the parties were in agreement on the point of interpretation Mexico had originally submitted, namely whether the review and reconsideration requirement constituted an obligation of result. Second, the two apparent differences of views seemed to concern an irrelevant question of fact (do we really need to know whether for example the governor of Texas shares the view of the US executive on the implementation of Avena? And even if he does not, what consequences could that have in international law?); and another question that may strike one as rather self-evident: of course, as Mexico stated in its own application and the United States accepted in oral arguments, all organs and subunits of the state are bound by its international obligations (though not all of them may speak for the state). Moreover this alleged dispute comes somewhat out of left field, as it did not form part of Mexico’s request for interpretation. Rather the federal and state authorities were only mentioned in the request for provisional measures (see the Joint Dissenting Opinion of Judges Owada, Tomka, and Keith).

To be fair, one has to mention that the Order only found the smallest majority possible, a 7:5 of the twelve sitting Judges, and that all five dissenters wrote or joined dissenting opinions – attacking the majority’s reasoning in no uncertain terms.

One point mentioned in the joint dissenting opinion of Judges Owada, Tomka, and Keith deserves in my view special attention. First, as the dissenters point out, it is problematic to allow a dispute in the sense of Article 60 between states to be based on the views of any territorial subunit or authority of one state. However, it is a second, in my view equally dangerous step not stressed in the dissent, to allow the inference from action or inaction of a state institution, to the conclusion that such state must interpret the obligation at issue as one of conduct, rather than result. Simply put, the argument goes like this: if Texas schedules the execution of Medellin, Texas/the United States must believe themselves not to be under an obligation to comply with the ICJ judgment which imposes an obligation of result. This logic would allow a state party to a case already adjudicated by the ICJ, in which an obligation of result was found to be upon another party, to bring a claim to the ICJ for interpretation of the judgment under Article 60, whenever the other state threatens to violate or has already violated the respective obligation of result. Even if some would hope differently, there is at present no basis for bringing such a claim against a state before the ICJ for failure to implement, as it were, a judgment of the Court.

Adding to the problems, all dissenters criticized that the order of provisional measures did not add anything beyond obligations the United States was already under by the terms of the Avena judgment, namely to provide review and reconsideration to all individuals identified in it. This of course precludes the possibility of executing an individual that has not yet been afforded such review and reconsideration.

All in all, the Order of Provisional Measures stirred up a lot of controversy, but sadly failed to achieve its main objective. Despite the ICJ’s Order of Provisional Measures, Mr. Medellin was executed by the state of Texas on August 5, 2008.

Having been unable to save the life of Mr. Medellin, the ICJ delivered its Judgment on the request for interpretation about five months later, on January 19, 2009. Contrary to the decision on provisional measures, the Judgment only surprised in one respect: it all but reversed the reasoning contained in the earlier Order of Provisional Measures. In the Judgment, the Court goes to some length to construe two “possible views” of whether there was actually a conflict between the parties. One view, highlighted in paragraphs 33 and 34 is that the words and deeds of different US authorities could form the basis of a conflict of views between Mexico and the United States as to whether Article 153(9) Avena is binding on the US and all its subunits:

33. According to Mexico, the United States, by word and deed, has contradicted its avowed acceptance of review and reconsideration as an obligation of result.  Reference is made to the choice of the United States Government not to appear at the Supreme Court hearings on Mr. Medellín’s petition for a stay of execution.  Mexico also points to the very tardy attempts to engage Congress in ensuring that all constituent elements do indeed act upon this obligation.
34. Further, Mexico contends that the Supreme Court found that the obligation within paragraph 153 (9) could not be directly enforced by the judiciary on the basis of a Presidential memorandum nor otherwise without intervention of the legislature.  In Mexico’s view, this necessarily means that the obligation is not really regarded as one of result – a viewpoint not shared by the United States.

The other view, presented in para. 38 of the judgment, is that Mexico did not meet its burden to demonstrate such a dispute. The Court in effect points out that Mexico’s emphasis on different federal and state authorities came, as I had mentioned earlier, a bit out of left field and that the Court would have to puzzle Mexico’s argument together for them:

…The Application made reference to a dispute about whether the obligation in paragraph 153 (9) of the Avena Judgment was one of result; the United States rapidly signaled its agreement that the obligation incumbent upon it was an obligation of result.  The matters emphasized by Mexico seemed particularly directed to the question of implementation by the United States of the obligations incumbent upon it as a consequence of the Avena Judgment.  The various passages in the further written explanations of Mexico of 17 September 2008, while referring to certain actions and statements of the constituent organs of the United States and perceived failures to act in certain regards by the federal government, nonetheless remain very non-specific as to what the claimed dispute precisely is.  Further, it is difficult to discern, save by inference, Mexico’s position regarding the existence of a dispute as to whether the obligation of result falls upon all state and federal authorities and as to whether they share an understanding that it does so fall.

The Court then proceeds to dismiss Mexico’s request, stating that “[t]he Parties’ different stated perspectives on the existence of a dispute reveal also different contentions as to whether paragraph 153 (9) of the Avena Judgment envisages that a direct effect is to be given to the obligation contained therein.” Here the Court thus seems to be saying that the only possible dispute to be derived from Mexico’s submissions would presuppose that the Court answer not only the trivial question whether the obligations arising out of Avena are binding on all authorities of the United States, but rather the more pertinent question whether such obligations are to be directly enforceable domestically.

This in turn nicely sets up the Courts’ finish. It goes on to interpret the “question underlying Mexicos’s Request for interpretation”  to concern “the general question of the effects of a judgment of the Court in the domestic legal order of the States parties to the case in which the judgment was delivered”. This question, however, was expressly left open in the Avena judgment, which allowed the United States as the Court now observed “to choose the means of implementation, not excluding the introduction within a reasonable time of appropriate legislation, if deemed necessary under domestic constitutional law.”
In turn, Mexico’s request for interpretation did in the view of the Court not concern the meaning or scope of the Avena Judgment. Hence the Court found that it lacked jurisdiction under Article 60.

It deserves mentioning that the Court of course underscored the continuing unconditionally binding nature of the Avena judgment, and found the United States in violation of its Order of Provisional measures as Mr. Medellin was executed contrary to it.

By now emphasizing the importance of the relation of any dispute to the interpretation of the respective judgment, the ICJ thus managed to avoid a potentially dangerous expansion of Article 60. As regards the somewhat meandering course the Court steered to finally arrive at this conclusion, it may be fair to speculate that faced with the looming certainty of Mr. Medellin’s death by execution, the uncertain possibility to somehow still find merit in Mexico’s request may have swayed some to grant the Order of Provisional Measures. In the end, however, it was legal certainty that prevailed over speculative interpretation.

Tadic redivivus – Are we using the wrong test?

In my opinion Tadic is not the law with respect to attribution for the purposes of state responsibility, and I do not have much hope this will change any time soon, if we take state practice and opinio iuris, as well as the ICJ’s specific rejection of the test in Bosnia Genocide into account. Of course there is always room for considerations de lege ferenda. I agree, if it were the law, attribution of PMSC conduct to the hiring state would be easier, given that the Tadic bar is lower. In turn, this would reduce the gap in state responsibility of the hiring state for acts of PMSC personnel, as compared to the responsibility that such state would incur for acts of its national soldiers.

Stretching Article 5 of the ILC State Responsibility Articles

Hannah Tonkin makes the compelling argument that especially once we take into account the Commentary to Article 5, we may be able to attribute the majority of guarding and protective services under this clause to the respective hiring state. First, a word of caution may be in order when relying on the Commentaries to the ILC Articles as a source of international law. Although one can make persuasive arguments that they constitute a valuable subsidiary source for interpretation, they should in my view not be taken without further argumentation to carry equal weight as the Articles they accompany.

Second, taking the Commentaries at face value and granting Hannah Tonkin’s argument about Article 5, it would in my mind not apply to guards who are off-duty when violations are committed.  These guards are still not acting in he requisite capacity at the time they engaged in the relevant conduct, and thus still fall through Article 5.

Third, even if we take into account the location of the activity, and the persons for the benefit of whom the functions are executed, the “contextual approach” remains rather complex to wade through. Ultimately I believe it will be very difficult to properly assess whether elements of governmental authority were exercised at the time. Let’s stick with the guarding and protection example in a theater of armed conflict – protection of military officials may be a convincing example, but how about protection of diplomats? Maybe they would also qualify. But would we be willing to extend the reach of the exercise of governmental authority  to the protection of an official of a private firm involved in the reconstruction effort, if the protection is arranged by the hiring state? Moreover, what is the “overall purpose” of guarding and protection in a war zone?

I do believe Hannah Tonkin raises important issues here, but I also believe that the inquiry necessary to attribute under Article 5 is to date still not very clearly mapped out. Moreover, even if we could agree on a clearer test for the exercise of governmental authority in this context, once we add the factual uncertainties inherent in any operation in an area of turmoil. I am not so sure whether it will thus truly be possible to attribute the majority of guarding and protection services provided on behalf of states, as Hannah Tonkin suggests.

A general duty to prevent?

Professor Cryer’s “devil’s advocate” remarks, regarding a broadening of states’ obligations are very thought-provoking. Yet, I am not convinced by his suggestion that recent developments in State responsibility may have led to broader responsibility for failure to prevent actions than the ILC Articles might suggest.

As a preliminary remark, I do not think that there is doubt today in the law of state responsibility, or specifically the ILC Articles, that the violation of a (positive) obligation to prevent gives rise to state responsibility, as expressed for example in Article 1 of the ILC articles. The positive obligation suggested by Professor Cryer seems in my mind very similar in nature to the positive obligations I treat in my piece (although derived from customary international law rather than IHL or Human Rights). As such, it is also subject to the problem regarding the indeterminacy of just how much diligence is “due diligence”. However, just to be clear, such a general duty, if it exists, would not offer a way to get around the rules of attribution, rather it is a completely separate inquiry.

To establish state responsibility, we still need to show that the state either

1) violated a negative obligation, as a person attributable to the state engaged in internationally unlawful conduct (be it an action or omission); or
2) the state failed to meet a positive obligation with respect to the conduct of persons whose conduct is was at the time not attributable to it (e.g. an obligation to prevent)

Unless the question arises whether the conduct of individuals who may have been under a positive obligation mentioned in 2) was actually attributable to the state at the time, this form of state responsibility is independent from the rules of attribution (i.e. the conduct triggering the positive obligation does not have to be attributable to the state).  Of course, there are situations where nevertheless questions of attribution arise. One example would be a scenario where it remains unclear whether individuals who were seen at the scene of a massacre and didn’t do anything to stop it, were actually Ruritanian troops (presumably giving rise to responsibility of Ruritania), or rather frightened civilians.

Regarding Professor Cryer’s Article 16 argument, can Article 16 somehow become a standard of attribution? This would be quite contrary to the ICJ’s jurisprudence under Nicaragua. There is no rule of attribution based on aid or assistance, and knowledge of an offense is in my mind irrelevant to the question of attribution altogether.
Of course, knowledge could be relevant if it would be the trigger for a positive obligation, as discussed by the ECtHR in Osman and Mahmut Kaya.  Similarly, payment to the PMSC could constitute a trigger for a general positive obligation to prevent, however, I do not believe that international law as it stands can be interpreted that way.

Specifically, I am curious how the  Nicaragua judgment could support such a positive obligation. Granted, the United States was held responsible for “its obligation under customary international law not to intervene in the affairs of another State” (Merits, Judgment of 27 June 1986, operative para 3), but this to me seems to have been a violation of a negative obligation (not to interfere) and not a positive one, such as an obligation to prevent rebels from engaging in certain conduct.

As regards the issue of state responsibility in occupied territory, however, I completely agree with Professor Cryer that the occupying power has far-reaching obligations that can give rise to responsibility.

Summing up, it is the uncertainty inherent in approaches such as the ones suggested by Hannah Tonkin and Professor Cryer that leads me to believe that rather than seeking to attribute under Article 5, or putting too much faith into positive obligations, attribution under Article 3 HC IV 1907 and Article 91 AP I remains much more promising in international law as it stands today.

Editors note: Carsten Hoppe is on the Project Management team of the Priv-War Project  and Executive Director of the European Society of International Law. He is currently writing a PhD, on the same topic as his article, at the European University Institute. He has a JD from the University of Michigan Law School and has previously clerked for Judge Bruno Simma at the International Court of Justice. His article “Passing the Buck: State Responsibility for Private Military Companies”, (2008 )19 EJIL 989 -1014 is available  here.

What makes the video below from the Iraq war more disturbing than the countless others we have seen?

First off, Elvis Presley. The scenes of bullets flying into civilian vehicles are cut to a trivializing soundtrack of “Mystery Train” by the King of Rock ’n Roll. But what may be even more disturbing is that the video does not show a single soldier. Rather, it is attributed to a UK Private Military Company under contract with the United States, an employee of which claims to have shot the video.  Granted, it will be very difficult to assess who is ultimately responsible for the shootings. But this problem is not the end, but rather the beginning of my inquiry. How can it be that private company employees get to drive around in a war zone with assault rifles and machine guns, clearing the road of suspected attackers in approaching vehicles? Is it true that states hiring these Private Military or Security firms [PMSCs / contractors] can outsource their international responsibility by way of a simple contract?

Of course, at present, states are free under international law to outsource functions in armed conflict, such as guarding and protection, interrogation, or even combat, which formerly were in the exclusive domain of soldiers. However, the article and my broader work on PMSCs for the Priv-War Project and my dissertation demonstrate that, while they may spend the money, they are not free to “pass the buck” with respect to responsibility. Thus the most important conclusion of my piece is that where contractors function as armed groups and are responsible to the party through their obligations under a contract, responsibility for all their acts, as first envisaged by states in 1907, will lie for the hiring state.

To get to this point, I first compare responsibility of a state for a classical soldier to all the options for attribution of private conduct. In this analysis, a responsibility gap becomes evident: unless a state outright incorporates the contracted personnel into its armed forces, or the contractors can be regarded as completely dependent on the state (a tough burden of proof to meet), the state will always face less responsibility for acts of those persons than for acts of soldiers, and its responsibility will be harder to prove.

In a further step, I show that positive obligations of the state under IHL narrow this gap to some degree. The gap closes in international armed conflict with respect to interrogation contractors in POW camps, but the off-duty conduct of combat and guarding and protection contractors would still be checked only by the general duties to vet, train, instruct, and report, and possibly to prevent known ongoing violations. In occupation, the off-duty conduct of contractors providing coercive services may give rise to responsibility of the hiring state where it failed to exercise due diligence in supervising them. In non-international armed conflict, only the general duties to vet, train, instruct, and report could narrow the gap, exposing the state to a substantially lower responsibility risk as compared to the conduct of its national soldiers.

Next I introduce positive obligations under Human Rights Law [HRL]. Adducing positive obligations under HRL to equalize responsibility for contracted personnel exercising combat, guarding and protection, or interrogation services with responsibility for the states’ own soldiers is subject to a twofold limitation: first, limits on the extraterritorial application of the HRL instruments and, secondly, the due diligence nature of the obligations.

Extraterritorial applicability will have to be tested on a case-by-case basis. In the case where the responsibility gap arises, that is where the conduct of contractors is not attributable to the hiring state, the requisite control will have to be exercised by the national armed forces or other state agents, including contractors whose conduct is at the time attributable to the hiring state. Territorial control over the area in which the violation happened may exist by virtue of an occupation or other territorial control, for example during an invasion. Physical control on the other hand, putting individuals in the power of the hiring state, exists for example over individuals who are kidnapped or arrested in an impromptu fashion or detained more formally in a detention facility. All three instruments discussed (ICCPR, ECHR, ACHR) will apply where the hiring state is an occupying power, or where the violations occur in an area controlled by agents of the hiring state. However, where combat contractors are off duty, or guarding and protection contractors engage in conduct outside their instructed duties, and the area they operate in is not controlled by the hiring state, there will not be a basis for extraterritorial jurisdiction unless there is physical control over the victims. Where the victims at the time of the violation either are under the control of the hiring state or have been handed from hiring state control into contractor control, jurisdiction can also be established under all three systems. However, where the hiring state does not have control over the persons interrogated, the off-duty conduct of these contractors will not be within the reach of any of the three instruments I discussed. If applicable, the reach of the duty to prevent under the different instruments varies, as described in more detail in section 3 B of the piece.

Positive obligations of states with respect to the conduct not attributable to them that I discuss under the heading of the obligation to prevent human rights abuses constitute obligations of due diligence. Thus, even if extraterritorial jurisdiction can be established, the hiring state will not be held responsible if it can demonstrate that it exercised due diligence with respect to the contractors’ conduct. Positive obligations of hiring states with respect to conduct not attributable to them are of course important, and may in the specific situations where they are applicable contribute to ensuring that a hiring state will not effectively circumvent responsibility it would incur for soldiers by relying on contractors. Yet, bringing responsibility to bear will be much more difficult due to a multitude of factors, including the limited reach of positive IHL obligations in non-international armed conflict, the uncertainty inherent in the concept of due diligence, and the complex questions extraterritorial application poses in the HRL systems. Thus responsibility qua positive obligations is very far from the simple and effective responsibility rule that states envisaged for armed conflict when they first codified state responsibility for violations of IHL by state forces in 1907.

Thus, still finding the responsibility gap to open up in many cases, I suggest in the remainder of the article that certain contractors exercising coercive functions do not fall into the category of persons accompanying the armed forces. Rather, they can indeed be attributed to the hiring state as members of the armed forces. Where they are organized as a recognizable group and are expected to ‘shoot back’ beyond self-defence on a routine basis, the hiring state will be responsible under Article 3 of HC IV and Article 91 of AP I. Hence, the clearest case will be that of combat contractors, while certain convoy or personal protection contractors may qualify provided they operate as a recognizable independent group. The interrogation example seems most doubtful, especially where the situations resembles more the close integration of a small number of individuals into an operation run by the national armed forces. I furthermore demonstrate that neither the requirement that they be under a command responsible to a party, nor the often adduced criteria for obtaining POW status, ultimately exclude this classification.