The Unpredictable Dictators

Until the week before it happened, most people refused to believe that Russia would attack Ukraine. Despite repeated warnings from the Biden administration and widespread evidence that Moscow’s troops were massing on Ukraine’s borders, it was difficult to accept that Russian President Vladimir Putin would try conquering Europe’s largest state. “He won’t be initiating an escalation,” said French President Emmanuel Macron on February 8, just 16 days before the invasion. Ukrainian President Volodymyr Zelensky was also caught off-guard, saying at the end of January that Biden’s claims of a coming invasion were simply “panic.” The German government was so convinced Russia would not attack that its chief intelligence official was in Kyiv on the day the war began and had to be whisked out by German security personnel.

The invasion of Ukraine is not the first time that officials incorrectly dismissed warnings that a state would strike its neighbor. In 1973, Israeli policymakers rejected reports that Egyptian President Anwar Sadat planned to attack the Sinai, citing the fact that his air force could not strike deep behind their lines. In 1979, U.S. President Jimmy Carter did not believe Chinese leader Deng Xiaoping’s warning that China might invade Vietnam because Deng’s statement did not comport with Carter’s worldview. And until the 1991 Iraqi invasion of Kuwait actually began, the United States was convinced that President Saddam Hussein would not attack, even though facts on the ground indicated otherwise.

There is a reason officials fail to anticipate foreign misadventures. Policymakers and analysts typically use a “rational actor model” to make predictions, and in keeping with its name, the model holds that policymakers will act rationally. It predicts that leaders will pursue defined goals after carefully searching for all available information and weighing the costs of different actions. But people are prone to make mistakes, and so this model is of limited use when forecasting what governments will do. It does an especially poor job at predicting the behavior of autocrats, who can pursue illogical ideas without domestic pushback.

This insight holds important implications for the way the United States and other democracies think about how to confront their adversaries. It is particularly critical for policymakers considering Beijing’s plans for Taiwan. It is unlikely that China has the military capabilities needed to take the island, which would require carrying out the largest amphibious operation in history. As a result, most analysts tend to believe an invasion is unlikely anytime soon. But this line of thinking assumes that Chinese leader Xi Jinping knows it would be impossible to seize and hold Taiwan without paying an enormously high price. In other words, it assumes that Xi is a rational actor when, in reality, he may not be.

Instead, surrounded by supplicants, Xi could persuade himself that a war for Taiwan would be fast. He could believe, as Putin did with Ukrainians, that Chinese troops would be welcomed by many Taiwanese people. He could decide that neither the United States nor its allies would come to the island’s defense. These assumptions are plainly wrong, but Xi would not be the first leader to make decisions that are disastrously incorrect. Washington, then, needs to be ready for a Chinese attack on Taiwan—even if it defies common sense.

RATIONALITY AND REALITY

It is easy to see why analysts are drawn to the rational actor model. What states do to each other can have tremendous consequences for millions of people. Leaders’ choices can also reshape the overall contours of the international system. Given these stakes, one would certainly like to think that—before making major decisions—leaders weigh the costs and benefits.

But in a world where decisions are made by individuals, rationality has its limits. Leaders, for example, do not usually consider all aspects of a decision they face. They struggle to make the calculations necessary to review the costs and benefits of all possible options. And people have difficulty determining what factors are relevant in any given situation.

The rational actor model also assumes there are universal, objective criteria that policymakers use to make choices; in fact, there are not. Different leaders have different priorities, and they pay attention to different pieces of data. Consider, for instance, the Cuban missile crisis. As the political scientist Jonathan Renshon found, U.S. policymakers failed to anticipate the Soviet Union’s decision to station nuclear weapons in Cuba because they did not put themselves in Soviet Premier Nikita Khrushchev’s shoes. Instead of fully appreciating the benefits that Moscow might gain—namely, the military and psychological advantages of stationing the missiles just off the coast of Florida’s panhandle—American policymakers focused only on the immense costs and high level of risk, as if Khrushchev viewed the situation in the way they did.

The Iraqi invasion of Kuwait provides another case in point. In the summer of 1989, a U.S. National Intelligence Estimate found that, although Saddam might flex his muscles, the Iraqi leader would not attack Kuwait because he would be rebuilding his military after the Iran-Iraq War. Saddam, the estimate argued, would prudently focus on paying back the immense debt he incurred fighting Iran, rather than spending money on a conflict of choice. But from Saddam’s perspective, seizing Kuwait’s lucrative oil resources was the only way to pay off his debts and preserve his regime. And his recent win against Iran, where the Arab world rallied behind him, left Saddam confident that he could annex Kuwait without much external opposition.

There are not universal, objective criteria that policymakers use to make choices.

To better account for adversaries’ perspectives, political scientists have created behavioral models that attempt to calculate how different countries perceive the world. In doing so, researchers have found (not surprisingly) that personality traits and emotions affect decision making in multiple ways. For example, self-monitoring—the ability to track and regulate one’s own emotions and behaviors—makes some national leaders more likely than others to fight over a desire to appear resolute. Leaders prone to defensive avoidance, characterized by efforts to evade or dismiss information that would increase anxiety and fear, often ignore distressing information (such as news about an impending invasion). Feelings can also shape how leaders interpret threats and when they decide to take action. During a crisis, for instance, emotions can make leaders either more reckless or more conservative than a rational actor model would predict.

Analyzing emotions can help analysts forecast the future. Such an approach, for example, might have produced a more accurate prediction of Khrushchev’s intentions during the Cuban missile crisis or Saddam’s motivation in the lead-up to his invasion of Kuwait. But even if analysts try to understand situations from an adversary’s point of view, they might make bad forecasts. It is hard to determine how much influence emotions have on a leader’s decision—and whether they pull leaders in a hawkish or dovish direction. Fear, for example, could prompt one leader to flee a dangerous situation while pushing another to fight. The same emotion can even have different effects on the same individual at different times. The reality, then, is that no model—however complex—can truly predict a leader’s actions.

Models fail at forecasting how all kinds of leaders will behave. But they are especially bad at predicting the actions of autocrats. Unlike in democracies, where the political process includes checks and balances that can stop bad decisions, authoritarian regimes have very limited, if any, checks on their leaders. Often, dictators ensconce themselves in an echo chamber that shields them from even hearing dissenting views. In Putin’s case, it appears that only a few top officials knew about his plans to invade Ukraine, and they all shared his beliefs and prejudices about Russia’s chances. In fact, Putin and his generals were so certain of a quick victory that when they invaded, soldiers were told to pack dress uniforms so a victory parade could be held in Kyiv.

EXPECT THE UNEXPECTED

Thankfully, there are ways for foreign-policy officials to account for uncertainty. The first is to game out the universe of mistakes that an adversary could make, consider the range of potential miscalculations, and then prepare various responses. To assess whether Russia might attack a NATO state, for example, analysts could map out the various ways Putin could expand its war beyond Ukraine. Then, they would evaluate the likelihood of each of these actions and consider what miscalculations would lead Putin to take them. Finally, analysts would generate a range of possible Western responses.

Even the best practices for structuring uncertainty, of course, cannot tell analysts precisely when adversaries will err. As a result, politicians need the intelligence community to keep close tabs on unexpected behavior. The community is well suited to this task; intelligence analysts specialize in looking for indicators that a dictator is about to make a mistake, such as military orders that contradict predictions, or indications that an adversary is mobilizing forces even when doing so does not seem wise. Human and signals intelligence on autocratic leaders can therefore serve as a kind of early warning system, allowing policymakers to expect foolish attacks.

The U.S. intelligence community performed this function in 2021 and 2022, chronicling Russia’s preparations for its ill-fated invasion. In the years ahead, it may have to do so again in the Taiwan Strait. Policymakers might think that Xi will steer clear of war because of how devastating an invasion would be for China’s own people, let alone for the broader region. But they need to understand that, depending on his psychology and assessments in the moment, Xi’s thinking about the island might depart from the rest of the Chinese Communist Party. It certainly could depart from what the West would think is a rational plan.

To divine Xi’s intentions, analysts must keep close track of China’s armed forces and economic plans. Large military buildups and moving troops into offensive positions, for instance, would be an obvious sign that Xi is considering an attack on Taiwan. So would Chinese efforts to increase oil reserves or to stockpile huge quantities of food. And analysts must keep close tabs on psychological indicators, from Xi’s speeches to whatever they can glean about the leader from human intelligence, to see what he is thinking. Leaders in the United States and its democratic allies must then pay attention to such findings so that they will not be caught unprepared. Western policymakers need to bake potential miscalculations into their analysis and commit to being open-minded, even if doing so challenges their worldview.

Ultimately, experts must remember that when it comes to dealing with autocrats, there are no certainties. Leaders will frequently ignore advice given to them or overestimate their abilities and miscalculate risks. In other words, leaders will not always be guided by level-headed rationalism, regardless of what outside observers think. It is a mistake for intelligence analysts and policymakers to assume otherwise.