Since Russia’s invasion of Ukraine, a paradox about escalation has emerged. The West carefully avoids certain kinds of involvement—such as sending Kyiv MiG fighter jets, setting up no-fly zones, and putting boots on the ground—for fear that it will provoke a greater war with Moscow. But Western countries do supply Ukraine with sophisticated artillery and intelligence targeting Russian officers and ships. They have sent intelligence personnel and special forces to Ukraine to share information and move military equipment within the country. The distinctions between these kinds of assistance can seem arbitrary and change over time. Yet those differences are taken seriously by both Russia and the West, and they have helped stop the war from spreading.

In Ukraine, the two most important limits are clear: the West has not directly attacked Russian forces, and every party keeps their operations confined to Ukrainian territory. Yet such boundaries are hardly the only ones at play. NATO, for example, has refrained from involvement that would fall safely within these limits, such as providing jets or organizing volunteer units, because Moscow might see such assistance as provocative. Russia’s response would likely also remain within those proxy and geographic limits. But such caution on the part of NATO is sensible, because a harsh Russian reaction could harm Ukraine’s civilians and its government, as well as the West, in new ways.

The complexity surrounding what lines NATO, Russia, and Ukraine are willing to respect—and which ones they are not—reflects the fact that the rules of limited war are messy. In Ukraine and other conflicts, escalation is an intricate dance, informed by history, geography, and universal distinctions between different kinds of wartime conduct. Both sides feel out what the other will tolerate, usually converging on a shared understanding of what is fair game and what is not.

To keep conflicts limited, then, warring parties must gradually test each other’s boundaries, as the West has wisely done. By moving slowly, the United States and its allies have managed to help defend the Ukrainians while gauging Russia’s tolerance. NATO should continue to up its involvement at only a gradual pace, figuring out Russia’s redlines by carefully watching how the country responds to Western moves. A deniable Russian drone strike in Poland, for example, could be a sign that NATO has pushed Moscow too far, in which case it should pull back. Above all, NATO must continue to obey the clearest of redlines, fighting only through Ukrainian forces and the keeping actual combat to Ukrainian territory. Otherwise, it risks a far more dangerous conflict.

SETTING BOUNDARIES

In one sense, the war in Ukraine hardly qualifies as limited. Few of the people who fled their homes in the wake of Russia’s invasion would see it as such. But the size and scope of a conflict is relative, and the scale of human suffering would be larger if the war spread to new territory or if combatants used new weapons. To stop this conflict from spiraling, it is critical that its participants grasp how their enemies think about escalation.

Thankfully, when governments try to feel out their adversaries’ limits, they aren’t fumbling in the dark. Writing during the early Cold War, the game theorist Thomas Schelling observed that limits in war—and thus what is considered escalatory—derive from crisp differences. Rivers, mountains, lines of latitude, political borders, uniforms, and weapons define the contours of escalation. During war, what counts as escalatory must be sorted out on the fly, but governments do not start from scratch.

Geography, for instance, offers guidance. During the Korean War, the twists of the Yalu River, the waterway separating North Korea and China, were a powerful marker. Bombing on the wrong side of the Yalu was seen by U.S., Chinese, and Soviet leaders alike as highly escalatory. In Ukraine, the Dnieper River, the borders with Poland and Russia, the location of the capital, and the distinction between eastern and western Ukraine similarly help define which military operations fall within the war’s established limits.

Other context-specific features can influence ideas of escalation, including culture, racial stereotypes, and language. In the Korean War, U.S. officials suspected that the Soviet Union might send volunteer units from communist bloc countries. A 1951 telegram from Alan Kirk, the U.S. ambassador in Moscow, noted that, if this happened, Soviet leaders might use personnel from “an Oriental satellite,” such as Mongolia, to be more deniable and less provocative—the implication being that European soldiers would have been more clearly foreign and thus escalatory than Asian ones. The Soviets never took this step, but the report showed how historically specific conceptions of racial similarity can inform what is considered escalatory.

Escalation is also shaped by universal ideas, many of which are simple distinctions in the laws of war. A country’s shift from bombing only military sites to civilian ones, or from factories to hospitals, is clearly escalatory. And some weapons are more provocative than others. In the 1980s, when U.S. leaders covertly supplied the mujahideen in Afghanistan to fight the Soviets, for example, they initially sent Swiss-made shoulder-fired missiles, believing that Moscow would react more harshly if they sent the U.S.-made Stinger system. Eventually, after observing the Soviet reaction, Washington did send Stinger missiles.

But when countries cross important limits, their adversaries do not always retaliate. In the first weeks of Moscow’s invasion, Western governments declined to send air defense systems, instead opting for smaller antitank ones. Some states refused to send lethal aid at all. But the West rapidly changed its mind. Within a month, the United States and the United Kingdom sent man-portable air defense equipment. Soon after, NATO countries began sending powerful artillery systems, including the U.S.-made High Mobility Artillery Rocket Systems (HIMARS), which can hit targets behind Russian lines. These shifts are easily recognizable as escalatory: assistance that shoots down planes is different from aid that stops tanks, and long-range artillery is clearly more aggressive than the short-range kind. Yet Russia has tolerated many of these gradual increases in lethality. One reason may be that such assistance still falls short of violating the war’s most important limits.

Russia’s lack of response shows that although the boundaries of escalation may be crisp, they aren’t always fixed: what is considered provocative in war can change dramatically. Early in the Vietnam War, for instance, the United States bombed North Vietnam sparingly, because doing so was seen as potentially escalatory. Gradually, however, such operations became routine as Washington realized it would not incite Soviet or Chinese retaliation. American targets crept northward. Toward the end of the war, U.S. President Richard Nixon even looked for targets that had been considered off-limits, such as ones in Cambodia. His goal was to gain leverage during peace talks by showing the North Vietnamese he would go where his predecessors would not.

THE LIMITS OF LIMITED WAR

Russia, Ukraine, and the West have similarly adjusted their behavior as they learn about the goals and pain thresholds of the other side. Escalation is not infinitely flexible, however. Some limits are more important than others. To prevent a wider war, NATO and Russia have carefully avoided direct, sustained military clashes between their own personnel. This is why proposals for no-fly zones are different from those calling for deadlier weapons. Although opinions differ about how aggressively to arm Ukraine, Western analysts agree that intervention must run through Ukrainians.

Ukraine’s borders are another universally understood limit. Even if they are disputed by Russia, they provide a powerful and intuitive way to contain the war. In March, Russia fired missiles near the Polish border but stopped short of targeting supply routes in the territory of a NATO member. Such restraint did not go unnoticed. Russia continues to avoid striking NATO members, despite having the capability.

These two lines are political and define the most explosive forms of escalation. A direct clash between NATO and Russia in the skies over Ukraine would invite a tit-for-tat response, fully upending the war as a proxy one and pressuring leaders on both sides to directly attack. Sustained military strikes by Russia into Poland would trigger NATO’s mutual-defense promise. If Ukraine openly attacked Russian territory, Moscow would feel compelled to respond in new ways. These transgressions would be distinct and ominous.

An exception to these limits is deniable operations. During the Korean War, Soviet pilots covertly flew for China’s air force, and American leaders knew but stayed silent. In the Vietnam War, the U.S. Air Force secretly bombed Laos, and the Soviets only complained privately. States often hide activity during war that breaks bedrock limits. Rivals may tacitly agree to stay quiet about these transgressions to avoid a larger war. Moscow may have done so in response to the rumored Ukrainian sabotage of fuel depots and factories within Russia.

But only some kinds of operations can be kept secret, and carrying out too many such attacks is dangerous—especially in Ukraine. A wider war in Europe would ruin economies and unleash carnage on soldiers and civilians, and it could upend domestic stability for all involved. In the most horrific scenario, nuclear weapons might be used in war for the first time since 1945. There’s a reason that both the West and Russia want to limit the conflict.

Some might think that Western governments made a mistake by not sending Ukraine lethal aid at the beginning of the invasion. But because limits evolve during war, being cautious was the right move. Notions of escalation evolve through trial and error. What NATO gets away with today is different from what it could do in February 2022. The West’s evolving approach to Ukraine—with its gradual expansion in volume and lethality of aid—has allowed it to find the forms of escalation that matter to Russia.

NATO members can continue to ramp up the lethality of their assistance. Yet they should do so gradually. A patient approach allows the West to gauge Moscow’s reaction and watch for signs of new Russian retaliation. Consider the question of whether to send MiG jets to Ukraine. Although this form of aid would not explicitly violate the war’s most important boundaries, because the planes could be used to attack Russia’s territory, Moscow could consider it an act of escalation and respond by, say, launching missile strikes closer to NATO territory or enacting new cuts to Europe’s gas supply.

Much of escalation is murky territory. But by going slow, the West can tease out the ambiguities without starting World War III.