Deterrence Works

Sep 8, 2011

By Jonah Friedman

A few weeks ago, Gareth Cook wrote an article in the Boston Globe about the scholarly historical work of Tsuyoshi Hasegawa, a historian at the University of California, Santa Barbara. Hasegawa argues that the Second World War was not terminated, as is often believed, by the use of nuclear weapons on Hiroshima and Nagasaki, but by the Soviet Union’s entry into the Pacific theater of the war. This suggests that the Japanese leadership was not compelled to cease fighting even by the destruction of their cities. This observation has in turn led Cook, and others, to question the efficacy of nuclear deterrence, based as it is on the assumption that the threat of catastrophic damage to an enemy’s civilian population (in one way or another) will prevent war. While it may be true that Japan was not brought to the negotiating table by the bombing of its cities, this example does not have the dire implications for nuclear deterrence that it might suggest. Nuclear weapons represent (in some ways at least) a significant enough departure from past weapons to induce caution in states where conventional weapons cannot. Moreover, the Japanese experience is rooted in the historical context of the Second World War, and is therefore less constructive for present and future deterrence than it seems.

Cook’s article describes how Hasegawa’s work shows that the Japanese leadership, far from clinging to some unrealistic and fanatical hope for victory in the summer of 1945, was in fact making a rational calculation. Aware that they could still make an invasion very costly for the United States, the government of Japan hoped to persuade the Soviet Union (at this time not involved in the Pacific theater) to mediate between Tokyo and Washington. However, with the Soviet invasion of Manchuria on August 8, 1945, that prospect was lost. Hasegawa notes that the atomic bombing of Hiroshima on the previous day caused concern, and “a sense of urgency,” but was not sufficient in itself to compel the country to capitulate.

There is certainly much to be said for this argument, but it suffers from a significant drawback. As Kevin Drum has noted in a recent article, the fact that the atomic bombings and the Soviet attack happened in quick succession makes it almost impossible to determine which was the true cause of Japan’s decision to surrender. One is compelled to agree with Drum’s assessment that “probably they both played a role.” Even so, accepting that the atomic bombings were not the sole factor (or possibly even the primary factor) in Japan’s has caused some to question where this leaves nuclear deterrence theory.

Ward Wilson, in a 2008 article in the Nonproliferation Review, argues that the inability of city-busting attacks to bring about Japan’s surrender, along with other examples of ineffective attacks against civilians, points to a fundamental flaw in nuclear deterrence. From his perspective, the fundamental threat underlying nuclear deterrence is the destruction of cities and the deaths of large numbers of civilians (he gives several plausible reasons why this is ultimately true). He notes that during the Second World War, Germany suffered some 570,000 deaths to aerial bombing and that a single raid on Tokyo killed approximately 120,000 people and destroyed 14 square miles of the city. He points out, however, that neither Germany nor Japan surrendered as a result of these losses, and concludes that “it is difficult to argue, on the evidence, that city bombing mattered to Japan’s leadership.” He also argues (correctly), that the atomic bombings of Hiroshima and Nagasaki “merely extended what was already a ferocious campaign of city bombing and were generally within the parameters of destruction for these conventional attacks.”

While all of this is true, Wilson seems to be selective about how important historical context is. He accurately places the escalation to nuclear weapons within the context of unprecedented conventional attacks on cities and civilians, but ignores this context when considering what it means for deterrence. To the Japanese leadership at the time, the destruction of a city with one bomb was not terribly different from the destruction of a city with thousands. In a war where dozens of Japanese cities had been effectively destroyed, the leveling of one more could not bring the Japanese government to surrender. This is not the context in which we find ourselves today. The prospect of large-scale conventional bombings of cities is a remote one. As Kevin Drum points out:

“America in 1945 had an air force capable of leveling cities with conventional weaponry. We still do—though barely—but no other country in the world comes close. With an atomic bomb and a delivery vehicle, North Korea can threaten to destroy Seoul. Without it, they can't. And larger atomic states, like the US, India, Pakistan, and Russia, have the capacity to do more than just level a city or two. They can level entire countries… It's not 1945 anymore.”

Moreover, nuclear warfare (especially in its modern form, with missiles capable of hitting any spot on the globe in under an hour) is of a different quality from conventional wars of the past – even one as destructive as the Second World War. Conventional wars occur because one side predicts it can achieve a worthwhile victory at an affordable price, and the other side calculates that it has a reasonable chance of defending itself (or at least, that surrendering brings greater costs than fighting). Of course, no state would fight a war which it knew at the outset would end in defeat, but the myriad uncertainties which war and interstate relations entail means that conventional warfare is not simply a matter of adding up numbers of tanks, planes, etc.

The 19th century military theorist Carl von Clausewitz wrote about the endless number of small factors that may alter the course of conventional wars, and the uncertainty which this engenders. This uncertainty is greatly reduced, if not eliminated, by nuclear weapons and their modern delivery vehicles. Where armies can be halted and aircraft can be shot down, there is nothing yet which can offer protection from a massive strike by nuclear-tipped ballistic missiles. Clausewitz also noted that, even in defeat, a state subjected to conventional warfare can expect its losses to be limited, and argues that “the ultimate outcome of the war is not always regarded as final. The defeated state often considers the outcome merely as a transitory evil, for which a remedy may still be found in political conditions at some later date.” Kenneth Waltz sums up the impact of nuclear weapons when he writes in The Spread of Nuclear Weapons: A Debate Renewed that:

“Countries armed with conventional weapons go to war knowing that even in defeat their suffering will be limited. Calculations about nuclear war are made differently…If countries armed with nuclear weapons go to war with each other, they do so knowing that their suffering may be unlimited…In a conventional world, one is uncertain about winning or losing. In a nuclear world, one is uncertain about surviving and being annihilated.”

Wilson might argue that this is all well and good in theory, but that the historical record shows no evidence that nuclear weapons actually deter war. He claims that “there is little evidence that either the United States or the Soviet Union was ever on the brink of launching an aggressive war against the other,” and that it is therefore impossible to show that deterrence works if there was no imminent risk of war. Although Wilson may be correct that neither state had any premeditated plan to attack the other, the Cold War is rife with examples of crises in which they may have resorted to war had the fear of eventual nuclear escalation not introduced great caution into their calculations. It is hard to argue, for instance, that the true fear driving President Kennedy during the Cuban Missile Crisis was that of Soviet tanks or aircraft alone. Rather, the thought of nuclear war made both sides reluctant to fight. Wilson may be correct that there is little hard evidence which explicitly links the absence of U.S.-Russian conflict during the Cold War to a fear of nuclear retaliation, but it is always difficult (if not impossible) to prove the effectiveness of preventive measures. Yet, the evidence which does exist, along with a significant theoretical basis and strong historical correlations, suggest the efficacy of nuclear deterrence.

Still, Wilson notes (correctly) that nuclear weapons have not been successful in deterring conventional attacks on states which possess them. He points to examples such as the Syrian and Egyptian invasions of Israel during the 1973 Yom Kippur War, as well as the Falklands War of 1982. However, there are two things wrong with this argument. The first is that the idea that states with nuclear weapons are invulnerable to conventional attack is a misconception; rather, they protect states from existential threats, be they conventional or nuclear. They do not guarantee that a state will never be attacked. The second problem involves his examples. There is evidence which suggests that during the Yom Kippur War, Syria and Egypt did not press home their advantage to the extent that they could have, and that their aims were quite limited due to Israel’s possession of nuclear weapons. Regarding the Falklands War example, according to one source, British Prime Minister Margaret Thatcher threatened to use nuclear weapons against Argentina unless French President François Mitterrand handed over secret codes which would make the French-made Exocet missiles in the Argentinean arsenal “deaf and blind.”

However, for Wilson and others of the same persuasion, examples of nuclear weapons preventing or limiting war are not sufficient to show the efficacy of deterrence. As he writes, “the mutual caution of the Cold War is evidence that nuclear weapons are dangerous, not that they are effective weapons of war or useful for threatening.” Wilson goes on to provide a historical counterfactual involving biological weapons instead of nuclear ones, arguing that neither are militarily effective, and posits that “nuclear weapons are dangerous (and induce caution) without being particularly effective. The caution on both sides during the Cold War is not proof of the deterrent value of nuclear weapons.” But what is the difference? Is that not the salient characteristic of nuclear weapons – that they are extremely dangerous, and thereby deterrence-inducing?

The historical record of nuclear deterrence in preventing major power war appears to be quite good. It is important when evaluating this record not to attribute to nuclear weapons benefits which they are not meant to confer (such as protecting against all forms of aggression), or to take historical examples out of context. The context of the bombing of Japanese cities should not be understood to indicate that nuclear deterrence is ineffective. Rather, it reflects a given set of circumstances which no longer exist today, and which have been largely eliminated by modern nuclear weapons and their delivery vehicles. The caution induced by these weapons since 1945 has indeed been useful, even if they were not sufficient to bring the Second World War to an end. Nuclear deterrence can be a powerful and useful force for preventing and limiting interstate conflict, but both its ability to do so, as well as its limitations, must be well-understood.


Jonah Friedman is a Research Intern for the Project on Nuclear Issues. The views expressed above are his own and do not necessarily reflect those of the Project on Nuclear Issues or the Center for Strategic and International Studies.


Two points

I appreciate the fair and careful representation Jonah Friedman has given my views. Three things.

1) I believe the record of Cold War crises is more ambiguous than most people presume. Most people presume that the record of Cold War crises mostly supports the notion that nuclear deterrence works. I'm not so sure. Sometimes it appears to support the notion that nuclear deterrence works, sometimes not. Take the Cuban missile crisis. The Cuban missile crisis is usually described as an example of nuclear deterrence working. The Soviets, when presented with the danger of nuclear war, withdrew their missiles, after all. However, this overlooks a problem. If the risk of nuclear war restrains leaders from aggressive action, why didn't it restrain Kennedy from imposing the naval blockade on Cuba in the first place? Kennedy was clearly aware that he was risking nuclear war. They talked about it for a week beforehand. After the crisis he told Sorensen the risk of war was between 1/3 and 1/2. Yet he took the action that would bring on the risk of nuclear war anyway. There are other examples of nuclear deterrence failing. How could Stalin have risked blockading Berlin if the U.S. had a nuclear monopoly? And so on.

2) I've got a question related to the Yom Kippur war example. If the Yom Kippur case and the Falklands case show that nuclear deterrence only works when your existence is threatened - you can't use it to protect far flung islands or occupied territories, for example - how can extended deterrence possibly work?

3) If we're going to risk the survival of our country on nuclear deterrence (and those are the stakes - the survival of the United States), aren't we required to be certain that it works? Imagine that I give you a device to keep your house safe from burglary and attack. It's an explosive device about the size of a small refrigerator. "I call it the Q bomb," I tell you. "Put this sign on your front lawn and tell all your neighbors that you've got this Q bomb in your basement and everyone will leave your house alone. Your family will be perfectly safe." A couple of days later your wife is in the basement doing laundry and she hears the Q bomb making some strange rattling, clunking sounds. You call me. "This Q bomb isn't going to go off by accident is it? It's really safe, right?" you ask. "Sure," I reply. "It's entirely safe. . . . Probably." Now, put down the phone, go upstairs and explain to your wife that the Q bomb is really safe . . . probably.

Because the dangers associated with nuclear deterrence are so high, the standard of proof when evaluating whether it works or not must be correspondingly high. I'm not persuaded that the evidence proves nuclear deterrence with sufficient confidence.

Ward Wilson
Senior Fellow
James Martin Center for Nonproliferation Studies