The “Last Window of Opportunity” Test Is Not the Law — and Shouldn’t Be

Much has already been written about the legality of Israel and the US’s attacks on Iran’s nuclear facilities — Operation Rising Lion and Operation Midnight Hammer — including excellent critiques by Marko Milanovic (here and here) and Adil Haque (here). I don’t want to rehash their arguments, with which I entirely agree. Instead, I want to explain in this post why the so-called “last window of opportunity” test, which has been invoked by Mike Schmitt (here) and Nick Tsagourias (here) to justify Israel’s attack, is not the law — and should not be.

Is “Last Window of Opportunity” Lex Lata?

The right of self-defence has traditionally been understood to have a temporal component. The most restrictive view, which is based on the text of Art. 51 of the UN Charter, is that self-defence is available only in response to an armed attack that is already in progress (“if an armed attacks occurs”).  A more permissive view, which is based on (one reading of) customary international law, is that self-defence is also available in response to an armed attack that is temporally imminent — one in which the need to act in self-defence is, to quote the famous Caroline case, “instant, overwhelming, and leaving no choice of means, and no moment for deliberation.”

More recently, a few states and a number of Western scholars have argued that even the Caroline standard — often referred to as “pre-emptive” self-defence — is too limiting.  These states and scholars insist that, in light of the reality of modern weaponry and warfare, the concept of imminence no longer has any temporal component at all. Instead, regardless of when an armed attack might occur, states now have the right to act in self-defence as long as they are facing the “last window of opportunity” to prevent the attack. Here is Mike Schmitt, writing more than 20 years ago:

[I]t would be absurd to suggest that international law requires a State to “take the first hit” when it could effectively defend itself by acting preemptively. This being so, the correct standard for evaluating a preemptive operation must be whether or not it occurred during the last possible window of opportunity in the face of an attack that was almost certainly going to occur. Restated, it is appropriate and legal to employ force preemptively when the potential victim must immediately act to defend itself in a meaningful way and the potential aggressor has irrevocably committed itself to attack. This standard combines an exhaustion of remedies component with a requirement for a very high reasonable expectation of future attacks – an expectation that is much more than merely speculative.

Nearly all of the scholars who have written on Operation Rising Lion and Operation Midnight Hammer accept that the legality of those attacks depends on the whether “last window of opportunity” is currently lex lata (the law as it is) or lex ferenda (the law as it should be). Schmitt and Tsagourias insist that the test is lex lata; Milanovic and Haque reject that claim.

Milanovic and Haque are right. State practice and opinio juris can expand the customary right of self-defence beyond the limits imposed by Art. 51. But even if the Caroline test should be considered lex lata — which is itself debatable — there is no question that the “last window of opportunity” test lacks sufficient state practice and opinio juris to qualify as customary international law. As both Schmitt and Tsagourias acknowledge, only three states have ever claimed the right to act in self-defence when an armed attack is not temporally imminent but the window of opportunity to act is about to close: the UK, the US, and Australia. We can probably add Israel to that list, because — as Haque notes in his post — its letter to the Security Council following Operation Rising Lion, though not specifically invoking self-defence, describes the operation as “a last resort” taken in “the last window of opportunity to prevent Iran from acquiring nuclear weapons.”

So that is four states in favour of the “last window of opportunity” test. What about the other 190? Here is what Tsagourias says: “Even if the positions of these States may not fully represent the views of the international community as a whole, they should not be dismissed merely because other States have not articulated their own interpretations of imminence.” To begin with, that is not how the formation of customary international law works. If only three (or four) states support the “last window of opportunity” test, it is perfectly appropriate to dismiss their views unless the other 190 have acquiesced to them. (See Henry here for a good discussion of acquiescence.)

More importantly, though, it is abundantly clear that the vast majority of the other 190 states have not only spoken up, they have “articulated their own interpretations of imminence” that leaves no doubt they reject the “last window of opportunity” test. Indeed, the 120-member Non-Aligned Movement (NAM), which represents 62% of the world’s states, has repeatedly denounced any attempt to expand the right of self-defence beyond the restrictive position in Art. 51, which requires an armed attack to actually be in progress. Here are two paragraphs, for example, from NAM’s statement following the 2019 Ministerial Meeting of its Coordinating Bureau (emphasis mine):

39.2. The Ministers reiterated the basic principle of the UN Charter that all States shall refrain in their international relations from the threat or use of force against the territorial integrity or political independence of any State, or in any other manner NAM 2019/CoB/Doc.1 23 inconsistent with the purposes of the UN. The Movement stressed that the UN Charter contains sufficient provisions regarding the use of force to maintain and preserve international peace and security, and that achieving this goal by the Security Council should be strictly done in full conformity with the relevant Charter provisions…. In addition, and consistent with the practice of the UN and international law as pronounced by the ICJ, the Article 51 of the UN Charter is restrictive and should not be re-written or re-interpreted.

40.6. [The Ministers] [o]ppose and condemn any categorization of countries as good or evil based on unilateral and unjustified criteria, and the adoption of the doctrine of pre-emptive attack, including attack by nuclear weapons by certain States, which is inconsistent with international law, in particular the international legally-binding instruments concerning nuclear disarmament; and further oppose and condemn all military actions including aggressive joint military exercises, or deployment of nuclear strategic assets, or use of force or threat of use of force against the sovereignty, territorial integrity and independence of Non-Aligned Countries which constitute acts of aggression and blatant violations of the principles of the UN Charter, including non-interference in the internal affairs of States.

In short, at most four states have explicitly claimed that the “last window of opportunity” test is lawful, while 120 have explicitly claimed that it is not. Contrary to Mike Schmitt’s claim in his post on Operation Rising Lion, therefore, there has been no “sea change in the law governing anticipatory self-defense.” At most customary international law permits self-defence in response to armed attacks that are temporally imminent. It does not permit self-defence in response to armed attacks that are not temporally imminent, even if a threatened state believes it is about to lose the “last window of opportunity” to defend itself.

Should “Last Window of Opportunity” Be Lex Ferenda?

Although the “last window of opportunity” test is not currently lex lata, one day it might be. That is for states to decide — all states, not just the Western ones that routinely use force. Personally, I hope states will continue to reject the test, because it is simply too prone to abuse. Debates over the the proper scope of self-defence reflect an inherent tension between two competing values: the ability of states to prevent armed attacks from causing them harm, and the need to ensure that states defend themselves only against genuine armed attacks. Lex ferenda, a strong case can be made that the conventional right of self-defence in Art. 51, requiring an armed attack to be in progress, is too restrictive in an era of hypersonic weaponry. But an equally strong case can be made that rejecting a temporal understanding of imminence entirely goes too far in the other direction, because it makes good-faith mistakes concerning an “attacking” state’s intention too likely and pretextual uses of “self-defence” too easy to justify.

An adequate explanation of why that’s the case would require a law-review article. For now, let me illustrate the danger of the “last window of opportunity” test by contrasting two Articles of War posts written by my friend Mike Schmitt (with whom I often disagree but for whom I have great respect). The first post, published in February 2022, analysed Russia’s invasion of Ukraine. Rightly rejecting Russia’s claim that the invasion was a lawful act of pre-emptive self-defence, Schmitt wrote the following concerning the “last window of opportunity” test (emphasis mine):

Russia’s actions are at best an example of “preventive self-defense,” at worst, and most accurately, pure aggression. Preventive self-defense occurs when one of the three essential elements of the last window approach is missing: 1) capability to mount an armed attack; 2) a decision to launch an armed attack; and 3) the need to act now lest the opportunity to effectively defend oneself be lost. Preventive self-defense employs force to deprive an adversary of its future ability to mount an armed attack (which the Russians are successfully doing) in a situation that does not allow anticipatory self-defense. Such actions are unlawful.

The unlawfulness of Russia’s invasion did not turn on Ukraine’s ability to attack Russia; the problem for Russia was that Ukraine had not decided to attack Russia so there was no need for Russia to “pre-empt” that attack. But the post is still notable for Schmitt’s explicit insistence that an act of self-defence cannot be lawful unless a state not only wants to launch an armed attack but actually has the capability to do so.

Now fast-forward three years, to Schmitt’s second post — this one addressing Israel’s attack on Iran. Schmitt could not simply apply the “last window of opportunity” framework he had presented in the context of Russia’s invasion, for an obvious reason — no state, not even Israel or the US, has claimed that Iran had the capability of attacking Israel with nuclear weapons when Israel and the US attacked it. So if Schmitt had insisted on capability, he would have been forced to conclude that both Operation Rising Lion and Operation Midnight Hammer were preventive self-defence and unlawful.

So what happened? Schmitt relaxed his 2022 test (my emphasis):

The Israeli dilemma is that this development is taking place behind the curtain. Moreover, the process of building a nuclear weapon lends itself to quickly assembling one, again without Israel’s knowledge. And once that has been done, it could be rendered nearly impervious to attack by storing it underground; numerous Iranian facilities related to the nuclear program are already underground, making effective attacks against them very difficult. Again, given the risk posed by nuclear weapons, it is not unreasonable to extend the condition of capability to the likelihood of near-term acquisition of that capability, especially when it can be easily rendered immune to attack.

So now the capability of mounting an armed attack no longer requires the actual capability to mount an armed attack. It simply requires the capability of  having the actual capability to mount an armed attack at some point in the “near-term.” To say that is a significant relaxation of the “last window of opportunity” test is an understatement.

More importantly, it is a problematic relaxation — one that indicates why states (and scholars) should not endorse the “last window of opportunity” test: it is a test for self-defence that is so devoid of clear limiting principles and so malleable that it can be used to justify any “defensive” act, even clearly unlawful ones. How long is “near-term acquisition”? A week? A month? A year? How immune to attack must a weapon-development program be for actual capability to mount an armed attack to be irrelevant? Completely? Partially? Is immunity from attack a sliding scale, whereby a state that has limited ability to destroy hardened positions can lawfully act in self-defence much earlier than a state like the US, which possesses a dazzling array of bunker-busting weapons? Is “easily rendered” itself a sliding-scale, one that actually favours weaker states because they will find it much more difficult than powerful states to protect their weapons programs from attack? And what does the “likelihood” of near-term acquisition require? Balance of probabilities? Clear and convincing evidence? Is a greater likelihood required for non-nuclear weapons, given their less destructive capability? Or would any particularly destructive weapon justify requiring less pre-attack certainty that the weapon will, in fact, be developed?

There are no easy answers to any of these questions — which is precisely the point. Permitting self-defence in situations involving a temporally imminent armed attack makes it difficult enough to assess the legality of many ostensibly “defensive” acts, as the West’s so-called “war on terror” has shown. Eliminating temporality entirely elevates those problems to a completely different level.

So the “last window of opportunity” test is not the law. And that’s a good thing, too. In the best-case scenario, widespread adoption of the test will lead to states making potentially catastrophic errors concerning whether other states actually intended to attack them. In the worst-case scenario — and to my mind, given recent history, the far more likely one — widespread adoption of the test will encourage states to invoke it to create a pretextual veneer of legality for aggressive uses of force.

We have already seen two such knowing misuses of the “last window of opportunity” test in Operation Rising Lion and Operation Midnight Hammer. As Milanovic says, “despite all of the reasonable, colourable, plausible or whatever arguments being made, this is not a tough case…. We all know it’s illegal. Just like we did with Iraq in 2003 or Ukraine in 2022.” Legitimising the test, however well-intentioned, will inevitably lead to the right of self-defence being abused even more regularly. And that is something all states and scholars should oppose.

CODA: I want to briefly address Schmitt’s claim that “it would be absurd to suggest that international law requires a State to ‘take the first hit’ when it could effectively defend itself by acting preemptively.” I’m normatively sympathetic, at least insofar as we are talking about temporal imminence and not “last window of opportunity.” Rationality, however, is not a necessary condition of a rule qualifying as either conventional or customary international law. Many international-law rules are absurd, yet we do not question their legal status — much less re-interpret them in a manner we find more appealing. For better or for worse, absurdity is the price we pay for a state-based international order that is founded on sovereign equality.