My new game-theory model (being applied here for the first time) shows a number of ways the First World War could have been avoided. Tens of millions lost their lives because a handful of diplomats played their cards poorly. That is the essence of Greek tragedy in modern times. Before I address how the war might have been avoided, let me provide a brief background on the circumstances that led to it.
Taking a very broad and long view, it is evident that the eighteenth and nineteenth centuries were a time in which those who adopted more democratic forms of government and more capitalist economic modes, such as the Netherlands and England early on and France later, enjoyed burgeoning wealth and influence in the world. Monarchy seemed to be in decline.
Zooming in on the latter half of the nineteenth century, we see the unfolding struggle to control Europe’s destiny. Germany as we know it did not exist for most of the nineteenth century. Instead, modern-day Germany was divided into many princely states—Prussia, Saxony, Baden, Württemberg, and many others. Austria dominated German affairs.
All of this was to change with the rise of Otto von Bismarck as Prussia’s minister-president (they certainly went in for awkward titles). Bismarck built Germany into a European power. First he united Prussia with several smaller German princely states to fight the Seven Weeks’ War (1866), in which, to the surprise of most European leaders, he quickly and easily defeated Austria. This war marked the end of the post-Napoleonic Concert of Europe system of checks and balances that had been forged half a century earlier to ensure stability among the great powers of Europe (then consisting of Austria, England, France, Prussia, and Russia) and to prevent the rise of another Napoleon.
The Seven Weeks’ War revealed that Austria was much weaker than its status as a European great power implied. Desperate to maintain itself among the ranks of important states, the Austrian government agreed to a merger with Hungary that resulted in the creation of the Austro-Hungarian Empire. This was the same merger deal that the Austrians had rejected just before their 1866 defeat. The creation of Austria-Hungary helped keep Austria in the running as a great power, slowing but not reversing its declining political position. Just four years later, Bismarck went to war against France, defeating Napoleon III in the Franco-Prussian War of 1870—71. With France’s defeat, Bismarck succeeded in unifying the remaining German princely states, creating modern-day Germany. Whereas Austria had dominated the pre-1866 concept of Germany, it was now excluded, not to be reunited with the rest of Germany until Adolf Hitler—an Austrian by birth—rose to power about sixty years later. By 1871, Bismarck had established Germany as the rising power of Europe and helped France to join Austria (now Austria-Hungary) as a state in decline. This set the stage for the First World War.
Revolutions against monarchy and oligarchy were bubbling up everywhere. There was revolt in Russia in 1905, in Mexico in 1910, and in China in 1911. From the Austro-Hungarian point of view, the most threatening emerging nationalist challenge to monarchy came from the Balkans. There the Austro-Hungarians saw in the experience of the Ottoman Empire the foreshadowing of their own demise. The kingdom of Serbia tripled its territory as a result of the Balkan Wars (1912-13), becoming a magnet for Serbian nationalists. They wanted all of Serbia out of the Austro-Hungarian Empire. Those tensions burst to the surface on June 28, 1914, with the assassination in Sarajevo, today the capital of Bosnia and Herzegovina, of the prospective heir to the Austro-Hungarian throne, Archduke Franz Ferdinand.
The assassination prompted Austria-Hungary’s government to issue an ultimatum to the Serbian government: give up your sovereignty, or it’s war. The Serbs were not without friends, and of course they were reluctant to give up their hard-earned independence. It seems that the Austrians were counting on this. The diplomatic records of the day, now open to us, reveal that they chose to make demands that they were confident could not be accepted. Apparently, the Austro-Hungarian leaders wanted a little reputation-building war with Serbia.
Europe’s great powers chose sides in the dispute. Russia sided with the Serbs. Under the terms of the Triple Entente—an alliance between Russia, France, and England—France and England also chose to side with the Serbs. The Russian decision triggered a response from Germany, by then Austria-Hungary’s ally. Under the terms of their alliance with Austria, Germany backed Austria-Hungary. Their broader alliance ties meant a high likelihood of additional support from Romania, Turkey, and especially Italy. The Dual Alliance of Austria-Hungary and Germany had been expanded in 1882 to include the new European power of Italy, with the expanded alliance referred to as the Triple Alliance.
Fearing an aggressive move by Germany in defense of Austria-Hungary, the Russians mobilized. They intended, in game-theory terms, to send a signal that they were committed to Serbia’s defense. This prompted a similar mobilization by Germany. Much as in the prisoner’s dilemma game we discussed earlier, each side could see that conciliation was better than war, but they also could see that trusting their adversary to pursue a settlement was risky. And so they found the result that follows from the logic of the prisoner’s dilemma: they fought instead of settling. In a few short weeks, the conflict over Serbia escalated to involve all of the great powers on the Continent. World War I had begun. The little reputation-building war between Austria-Hungary and Serbia was not to be.
Shortly, I will apply my model to inquire what might have happened if the Triple Entente of Britain, France, and Russia had been more able in 1914, or, for that matter, if the Dual Alliance of Austria-Hungary and Germany had been more skillful. First, however, let’s pretend that there was a little army for hire of people with good math skills pounding out my calculations in 1914. What would they have predicted with no advantages of hindsight?
To address the 1914 crisis—not the fighting of the war, mind you, but the diplomatic run-up to war—I constructed inputs for the computer program that measure the degree to which each of the European countries, plus important non-Europeans including the United States and Japan, favored Serbia’s or Austria-Hungary’s foreign policy in 1914. I estimate salience based on a mix of expert judgments and geographic proximity to the Austro-Serbian crisis. Potential influence is based on a standard measure of “national power” collected for every country in the world for every year from 1816 to roughly the present by the academic enterprise, introduced earlier, called the Correlates of War Project. Of course, I use the estimates for 1914. Since I am examining this case using my latest model, I include the additional variable it requires. This variable measures the extent to which each player is resolute in the position it has taken even if that means a breakdown in negotiations or, conversely, is sufficiently eager for an agreement that it will show considerable flexibility in its approach to negotiations. Put in terms of our earlier discussion of health care, this new model includes an input that calibrates how much a player’s bargaining style looks like Bill Clinton’s (100 on this variable’s scale) or like Hillary Clinton’s (0 on this variable’s scale) back in the early 1990s. This “commitment” variable’s values are based on my reading of the historical record in the run-up to World War I. Any interested decision maker in 1914 would have had access to the information used here. And if they had my equations, they could have done the exact analysis I report.
Austria-Hungary, Germany, Romania, and Italy start off at position 100, indicating a full endorsement of Austria’s position against Serbia following Franz Ferdinand’s assassination. Serbia and Greece start off at a position of 0 on the issue scale, indicating their total opposition to Austro-Hungarian demands for Serbia to surrender its sovereignty. The rest of the European states fall approximately between 33 and 45 on the scale, suggesting that they tilted toward Serbia and against Austria but not decisively.
As seen in figure 9.1, a palace full of bearded mathematicians crunching away on the numbers in 1914 would have anticipated war. They also would have realized that war could be avoided if they just crunched the numbers long enough, reflecting a prolonged diplomatic eff
ort instead of a rush to war. The figure shows that the model anticipates war sometime in August 1914. This is the stage at which the model’s logic says diplomatic efforts to resolve the dispute without resorting to the use of force would have ended.
FIG. 9.1. The Predicted Failure of Negotiations During the 1914 Crisis
You see that the model is constantly calibrating the expected benefits from continuing to negotiate, weighing them against the model’s estimate of the expected costs from continuing to pursue diplomacy. Eventually, in the absence of an agreement, the players conclude that the prospects or value of a future agreement just isn’t worth the effort. In essence, the model’s algorithm makes a judgment about the value the players attach to extracting a concession tomorrow compared to extracting the same concession today. Getting some benefit sooner is always worth more than getting it later. In this instance, the period that corresponds approximately to early to mid-August happens to be the time the model says the game would end because too little progress was being made in closing the gap between Austrian demands and Serbian concessions. So the game predicts its own end in August. At that point there is no agreement between the main antagonists, and so, according to the model, a new game starts, with generals taking over from the diplomats.
Up to this point the Austrians (supported by their German allies) have persisted in demanding enforcement of the Austrian ultimatum. Meanwhile the Serbian government has gone a good distance toward meeting many of Austria’s demands. Still, Serbia shows no willingness to accept the Austrian ultimatum, exactly as the Austrians hoped. Instead, Serbia adopts a moderately conciliatory posture that is consistent with the concessions pushed by the British, French, and Russians. The latter three, according to the simulated crisis, were believed to be strongly committed to finding a settlement. Because of that, neither in reality nor in the simulation did the Austrians and Germans think their foes in the Triple Entente were likely to go to war on Serbia’s behalf.
What were the Italians, members of the Triple Alliance, up to during the crisis? In reality, they indicated on July 28, 1914, that they could not support the Austro-Hungarian ultimatum, delivered just five days earlier. With war imminent, the Italians declared themselves neutral. They resigned from the Triple Alliance on the grounds that Austria-Hungary was launching an aggressive, not a defensive, war.
In the model’s assessment, the Italians start out at the same position as the Germans and Austrians, befitting their membership in the Triple Alliance. As can be seen in the figure, by mid-July in model time the Italians break from the Triple Alliance and become neutral. They adopt a position hovering around 50 on the issue scale. So the model sees the Italians moving a week or two earlier than they actually did, but nevertheless it foresees their shift to a neutral position. In the model’s logic, and in reality, the Italians were not committed to standing by their Austrian and German allies once they recognized that the unfolding events were going to produce something vastly larger than a small Austrian-Serbian war.
The model indicates that Austria-Hungary and Germany expected war over Serbia for the first few weeks of the crisis, while Serbia shared that expectation. Austria actually declared war on Serbia at the end of July while professing to have no quarrel with others. By the start of August the model anticipates that Serbia was relegated to a relatively minor role as events ran ahead of Austria-Hungary and Germany’s ability to cope with them. Germany in fact declared war on Russia on August 1, and the war anticipated by the model’s logic actually began.
Was war inevitable? Emphatically the answer must be no! First, we can see in the figure that had the contending parties sustained negotiations one or two months longer, postponing the decision to go to war, the Germans would have better comprehended the dangerous big picture. They would have (according to the model’s predictive logic) broken ranks with the Austrians and come to agreement with the British and their allies. The Austrians would have gotten the majority of what they wanted under the agreement that the model indicates could have been reached in September or October 1914. Of course, this agreement needed the diplomats to remain in charge instead of turning choices over to the generals. The deal that could have been struck would not have included the surrender of Serbian sovereignty. But all that is beside the point since, alas, the game simulation suggests correctly that the diplomats would not have continued negotiating through the early fall. The diplomatic game ends before September and a new game, war, begins.
The beauty of a model is the freedom it gives us to ask lots of what-if questions. We can replay the World War I diplomacy game, just as I did in an earlier chapter for a litigation client, while changing how players present themselves. That way we can see if this or that player could have approached the game better, producing a happier result from its point of view.
Let’s replay the 1914 crisis, this time making the British diplomats more skillful than they actually were but no more skillful than they could have been. I am going to let them look inside the model’s approximation of what was going on in the heads of the German and Austrian decision makers. In this way I am going to pretend that they had a little army of mathematicians doing the calculations my computer does for me. This will make it easier for the British to be more thoughtful and decisive, instead of as wishy-washy as they were.
The historian Niall Ferguson has argued that a big factor leading to war in 1914 was that the Austrians and Germans were uncertain of British intentions and that this uncertainty was caused by the British.3 Britain may have done well for a long time by muddling through, but that was not much of a strategy in 1914. Did the British really intend to defend Serbia, or were they bluffing? Certainly nothing they said or did at that time was sufficient to convince the powers of the Dual Alliance that defending Serbia was really important to Britain. This was an important failing on their part, and it deserves further exploration.
Remember that when we looked at a lawsuit I worked on, we examined the consequences that followed when I advised my client to bluff having a stronger commitment to their bargaining position than in fact they had. Such a bluff can be risky and costly. If the other side believes—correctly—that a tough posture is just posturing and not the real thing, then they will call the bluff. In the lawsuit, that would have raised the odds of a costly outcome. The client would have faced severe felony charges. They might have been exonerated in court, but trials are, as we’ve seen, always risky business. Without bluffing, they were going to face those charges anyway, so bluffing looked (and proved) attractive.
Think how much costlier and riskier bluffing could have been for Britain in the summer of 1914 than it was for my client in the lawsuit. With hindsight we know that the guns of August were not stilled for more than four years. At the end of the war, the United States—not Britain, not France, not Germany, and not Russia—would be the greatest power in the world. At the end of the war, Austria-Hungary would not even exist. But when decisions had to be made, no one knew any of that. They had to think about what their circumstances would look like if they showed eagerness to compromise or if they showed real resolve to stick to their guns, so to speak. The British looked for compromise, and disaster followed. What does the model say would have happened had they bluffed being resolved to defend Serbia, and how could they have conveyed such resoluteness?
The British were in an odd position. It seems that even they were unsure how resolved they were. They were uncertain not only about others, but apparently even about themselves, about what they should or would do. That, presumably, is why the Austrians and the Germans did not read British diplomacy as signaling real commitment to defend Serbia. We also know that when the Russians—believing they were facing an imminent attack—mobilized, this prompted the Germans to do the same, and war began. The Russian mobilization certainly showed their commitment, but it did nothing to improve the prospects of a negotiated settlement. Their mobilization was a very costly “costly signal.” Would British mobilization have had the same dangerous
consequences, or could it have broken the impasse?
The data going into the model treat Britain as highly committed to finding terms that all sides can accept. They were assigned a value of 90 out of a possible 100 on “flexibility/commitment,” indicating they really wanted to negotiate and were prepared to live with a major compromise to avoid war. I have repeated my earlier simulation of the crisis, but with one change. I shifted Britain’s commitment to compromise from 90 to 50. A value of 50 signals a balanced approach. A value of 50 means the player actively pursues a settlement but is sufficiently resolved that it will not make a deal very far from its desired outcome. By placing Britain at 50 I am, in essence, trying to test Niall Ferguson’s insight (and that of other historians too) that the wishy-washy British message contributed to the war. I am simulating an approach that the British leaders probably would have seen as a bluff intended to shake up the situation and promote a war-avoiding deal.
What concrete actions might the British have taken to send the message, “We are serious about defending Serbia’s sovereignty”? I am not a military expert, so my speculation will be just that. I am sure a military specialist or historian of British policy in the run-up to World War I would find countless other ways for the British to send the right message. Here is one:
Britain was the world’s greatest sea power (although the Germans were certainly challenging that claim at the time). They could have filled several of their navy’s ships with a few thousand British troops to be transported to the Adriatic, taking them just a short distance from Serbia. Maybe they could have sent some other ships into the Bosporus, roughly flanking landlocked Serbia from either side. This would have served several potentially advantageous purposes. It is very much, in game-theory lingo, a costly signal. Talk is cheap, but sending a fleet into a prospective combat zone is putting your money where your mouth is.
The Predictioneer’s Game: Using the Logic of Brazen Self-Interest to See and Shape the Future Page 21