A look back on exam day (WWI in Real Time, Exam 1)

As this post goes live, my class is taking the first (of three) exams in World War I in Real Time, which is as good a time as any to take stock of what ground we’ve covered so far, though mostly from a teaching perspective; this might be a little too “inside baseball” if you’re usually here for the WWI stuff.

Thus far, we’ve had 8 lectures—certainly feels like there’ve been more—covering

  1. The Puzzle(s) of the Great War – a syllabus day, but one that allowed me to set up why the war itself is (still) so puzzling, so surprisingly modern, and so worthy of study for all it can teach us not just about war, but about politics more generally.
  2. Strategy, Equilibrium, and Tragedy – a tool-building day, introducing some of the building blocks of game theory, developing the students’ sense of thinking strategically (and why games are good for helping us do just that), and addressing a historical puzzle—the occurrence of the Anglo-German Naval Race—that helped set up our broader discussion of the causes of the war.
  3. Europe Before the Great War – a day devoted to the logic of inference, emphasizing the need to compare a crisis that does escalate to war (the July Crisis) to similar great power crises that didn’t in the years before the war; also a great chance to learn about costly signaling, where some states would intentionally limit military power to make their claims of peaceful intentions more credible.
  4. Battle Plans, Strategy, and Equilibrium – a day of jumping into the heads of military planners before the war (Schlieffen and Moltke, as well as others in each of the great powers), trying to understand their dilemmas (and general lack of anything resembling a good option) through the Colonel Blotto game, which also brought home how difficult it is to judge the wisdom of a course of action ex post when its success depends on the actions of the other side.
  5. Germans Aims in the July Crisis – the first day where we dug into the decisions that would turn an Austro-Serbian crisis into a world war, focused in particular on what Germany wanted out of the crisis; in terms of explaining the war, we tried to suss out Germany’s preference ordering, showing how several actions would be linked to Russian choices in response. In essence, we wrote down part of a game-theoretic model of the July Crisis.
  6. Russian Goals in the July Crisis – day two of building a model as a class, figuring out Russian goals and how they’d relate to German choices outlined in the previous class, before “solving” the game and showing that, while Germany got its second-best outcome, Russia got its third best. Generally, we came to understand the war as the result of a series of interlocking prisoner’s dilemma-style commitment problems that pushed the powers into preventive war.
  7. Public Opinion and the Outbreak of War – a puzzle-solving day, where I led the class through solving a mystery: if Russia and France were so committed to the offensive, why did both take steps to hamper their own offensive capabilities once war was imminent? The answer was the need to secure public support for a mass war, but getting there required working out tradeoffs and a pretty tough strategic problem; where I walked the class through his in pieces in lectures 3-6, we did it all at once in this lecture. (Might’ve been my favorite thus far.)
  8. Belgium and British Intervention – another puzzle-solving day, trying to figure out why an attack on France might not’ve sufficient to bring the British into the war but an attack on small, less powerful Belgium was. the answer has to do with the informative and coordinating effects of international law, particularly what it told the British about German goals in the war. also got to spend time talking about coordination problems in general, and assigned one of my own papers—turned out to be a good chance to refine my own argumentation.

Looking back at this, I’m encouraged, though as slow as it might’ve felt to the class that we didn’t get to the outbreak of the war until late September, I was nervous that I was going too quickly in terms of building up the intuition of game theory. I’ve also never tried to use so many diverse strategic models in a class with so little actual formality and math, but I’m guardedly optimistic about the approach—and using it again. I guess the exam, though, will be the arbiter of how well the pedagogical plan has worked.

As for me, I’ve already gotten a few research ideas out of the course—wheels are spinning about the public mobilization stuff, and I’ve pulled some great examples for a few different papers from the readings—so in that respect, this course is already paying dividends.

Until Thursday, when we spend some time on the battlefield..

Belgium and British Intervention (WWI in Real Time, Lecture 8)

Today, we tackle the puzzling sequence of events that bring the United Kingdom fully into the war against the Central Powers—something about which even its French allies were “profoundly uncertain” until the 11th hour—and that would help make the war, thanks to the involvement of the Dominions, properly global. Just as the Asquith government was divided over the wisdom of sending an army to the Continent, even as Austria declared war on Serbia and Russia mobilized, so was the British public. However, once Germany sent an ultimatum to neutral Belgium demanding passage for its troops on the way to France, British public opinion—as well as remaining fence-sitters in the cabinet—swiftly turned towards favoring intervention. Why then, when war had been imminent for some time—a war that would involve Britain’s most valuable Entente partner in France—and why over small, non-allied Belgium, in particular?

The answer we develop in class is built around some of my own recent work (an early version of the paper is discussed here), which argues that the international laws of neutrality, which are publicly known, can help third party states (like the British in 1914) identify useful interventions—but only when the laws are violated, as they were when German troops stormed through Belgium in an attempt to wheel around the bulk of the French army. Compliance, on the other hand, might allow expansionist states that it’s worth fighting to mask their intentions and prompt intervention, or balancing, only when it’s too late. In fact, I argue that we shouldn’t judge some laws—in particular neutrality laws—on rates of compliance when their workins are less about compliance and more about identifying profitable interventions.

Waverers in the UK did seem to view intervention as worthwhile once Germany tipped its hand; the violation of Belgian neutrality, which both Britain and Germany expected would provoke British intervention, thus served as a strong signal to the British leaders and public that German war aims were more expansive than a mere desire to support the Dual Monarchy in a punitive campaign against Serbia. Consider a counterfactual: had it only tried to engage and tie down the main French force on their common border, claims that Germany was only aiming to support the Dual Monarchy might’ve been more credible. As it was, going through Belgium led the British to believe that, in victory, Germany couldn’t credibly promise not to dominate the Continent—a belief that would be confirmed in many Allied minds with the publication of Germany’s “September Program,” which stated that “[t]he aim of the war is to provide us with [security] guarantees, from east to west, for the foreseeable future, through the enfeeblement of our adversaries” (quoted in Hastings, p.100). While this might’ve been a wishlist at the time, it was certainly a set of aims that the British expected that Germany would be unable to turn down if it were victorious on the Continent, with Belgium conquered and France prostrate.

In light of this commitment problem, which became acute once the British became more convinced that Germany would attempt to dominate the Continent, we can see

  1. how the violation of Belgian neutrality—as opposed to the imminent invasion of a bigger ally in France—tipped the scales in favor of British intervention, and
  2. how international law can work, helping states identify desirable interventions, not despite but because it is violated.

If you’re interested in the workings of the laws of neutrality model, I’ll post a new version soon. In the meantime, after Tuesday’s exam, we’ll dive straight into the homefront and the battlefield as our protagonists take in “the superb spectacle of the world bursting into flames”…

Public opinion and the outbreak of war (WWI in Real Time, Lecture 7)

After working through the strategic implications of German and Russian goals in the July Crisis over the course of the last week, particularly what it meant for explaining the outbreak of the Great War, we spent time today on another puzzle—one that had never really struck me until prepping for today’s class, in fact.* We noted a few weeks ago that the Continental armies were all focused on the offensive, on striking the first blow and achieving a decision that would keep the war short.** However, two things stand out about the maneuvering in the early days of the war that don’t fit neatly into the idea of the “cult of the offensive“:

  1. Germany delayed mobilization until it could claim that Russia had mobilized first.
  2. France mobilized, but orders its forces to stay 6 km back of the German border.

In each case, a belligerent that knows war to be imminent gives up something valuable—Germany a mobilization lead over the Entente Powers and France the ability to do anything other that mount a defense in the early days of the war—that would seem heretical to the ideal of offensive operations. Why?

The answer, in each case, turns out to be public opinion—specifically, the desire of both governments to ensure that as much of the public as possible would be on board with the war, whether as willing conscripts, temporarily silent dissidents, or factory workers making shells instead of striking. (Merely locking up the leaders of socialist or workers’ parties wasn’t going to cut it in this case.) Making the war a national one of defense or survival (said, surely, with a bit of a wink in the Kaiserreich) would be key to winning a long, drawn-out war, should it come to that—Germany, after all, knew that beating Russia would take a while—but there were also strategic reasons to secure public opinion and achieve a mass mobilization of soldiers and laborers: if any belligerent failed to mobilize the whole of the public, it would be disadvantaged at the outset, and if would certainly want to mobilize the whole public against an opponent that hadn’t done so in order to secure a quick victory. So, oddly, while the desire to capture public opinion resulted in both Germany and France trading off some of the benefits of a first-strike strategy, these intentional disadvantages might’ve canceled to a degree—and thus masked the true effect of mobilizing public opinion on success in war, which requires a disparity to realize. (From a game-theoretic perspective, the effects are both canceled out in equilibrium, in the behavioral prediction, but the reason this is so is the desire to blunt any advantage an opponent might have from securing more public support than oneself.)

So what can we learn from looking at public opinion in the outbreak of war?

First, even in a semi-autocratic state like the Kaiserreich, it drove governments to trade off some of the benefits of a first-strike, offensively-oriented strategy, in order to ensure that their opponents couldn’t secure a unilateral advantage in mobilizing the whole public for war. At a minimum, it confirms for us that governments made tradeoffs in the face of strategic dilemmas—that they weren’t necessarily the outmoded, single-minded wellsprings of outdated ideas that it’s easy to see them as.

Second, and perhaps more disquieting, by enabling the mobilization of the nation qua nation for war, governments maneuvering to secure public opinion might have helped create the very conditions that made the war as long as it was and gave to the war its brutal, extractive, resource-driven, attritional character. Absent some ability to commit not to make it a “people’s war,” one that mobilized nearly everyone in society, governments presiding over a conscious, nationalist mass public found themselves leaping headlong into a drawn-out cataclysm, the alternative to which was losing—and much sooner—to an enemy that did bring the nation along with it.

* Thanks, by the way, are in order to Phil Arena for helping me out with the ideas here:

** Contrary to the conventional wisdom, there weren’t many illusions about what would happen if decision weren’t achieved early; the strict focus on the offensive was designed to avert the long, bloody, total war that ultimately did ensue.

Russian goals in the July Crisis (WWI in Real Time, Lecture 6)

Today’s lecture was very much a companion to Tuesday’s (which laid out for us German preferences in the July Crisis), as we fleshed out the other side of the equation that would take the world from an Austro-Serbian war to a general conflict involving the other Great Powers: specifically, Russia.

We worked out Tuesday that Germany’s most-preferred outcome of the July Crisis appears to have been successful localization of an Austro-Serbian war, which would have led to either Serbia’s forcible realignment with the Central Powers or its dismemberment and distribution of its territories to ostensibly grateful nearby states—handing Germany something close to a preventive outcome against Russia (by more or less permanently shutting the latter out of the Balkans and the Straits while bolstering the Dual Monarchy) at no cost. However, given that Austria couldn’t commit not to crush Serbia, which the delay between assassination and ultimatum made hard to believe—to say nothing of word leaking out all over the Continent about the Dual Monarchy’s extensive aims—Russia, it seems, couldn’t commit not to get involved to save Serbia. However, Russia knew that it would be stronger by 1916, and surely seems to have preferred most of all that war be avoided. As such, we worked out this tough preference ranking over outcomes, from top to bottom:

  1. Austria backs down (or just stops in Belgrade, making it a clear punitive operation)
  2. Austria attacks Serbia, Russia intervenes, Germany stays out
  3. Austria attacks Serbia, Russia intervenes, Germany intervenes
  4. Austria attacks Serbia, Russia doesn’t intervene

Clearly, Russia was able to avoid her worst outcome—losing the Balkans, and the promise of access to the Turkish Straits, to the Central Powers—but nonetheless ended up with her third best: a general war involving the Great Powers. Russia was able to guarantee that Germany didn’t get its best option, but Germany was able to force Russia down to its third, in what looks like a grim series of commitment problems: neither Russia nor Germany could commit not to take advantage of the other’s standing pat (Russia because it couldn’t stand aside and see Serbia crushed, Germany because it now believed that Russia had turned the assertiveness corner and made preventive war necessary), and as a result we saw the beginning of a war that would ultimately drag in France, the UK, Turkey…and a great many others.

Ultimately, we can understand the escalation of the July Crisis as the result of three interlocked decisions for preventive war:

  1. Austria to solve her nationalities problem by ridding itself of growing, nationalist Serbia
  2. Russia to prevent the loss of its position in the Balkans to the Central powers by supporting Austria
  3. Germany to prevent Russia from completing the Great Program of rearmament and strangling the former’s growth as a Great Power

without any one of which the war might’ve looked much, much different than it did.

Still, at this point in the class—with Russia lurching from partial to full mobilization as Germany embarked on a mobilization of its own—the war was still regional. It won’t be until next week, when the French and British dominoes fall, that it starts to become a World War. What was important, pedagogically, was to nail down the logic of commitment problems and how they lead to tragic outcomes, and to force students to spend some time in the heads of decision-makers—which, when you’re trying to understand why a war occurred when and how it did, is no mean task—working out their preferences, available actions, and seeing what we can learn by taking their dilemmas, their choices, and their goals seriously.

German aims in the July Crisis (WWI in Real Time, Lecture 5)

After a week that saw us talk about prewar crises and military planning for the big one, should it ever break out, today we dived directly into the strategy of the July Crisis following the assassination of Franz Ferdinand. Rather than follow the crisis in a linear fashion, though, we tried to tackle a part of the story: German preferences over the many possible outcomes of the crisis. German goals in the war and the preceding crisis have long been a subject of debate—beginning almost as soon as the war broke out—and our goal was to use the weight of available evidence to figure out how Germany might’ve ranked the possible outcomes of its choice over backing Austria in its proposed war on Serbia; doing so will give us some tools to think about the preferences and choices of the other great powers (particularly France and Russia) when we pick up the other part of the narrative—or, more properly, flesh out the other side of the game—on Thursday.

Without rehashing the historiography, here’s what we settled on. Given Austria’s desire to crush Serbia—either dismembering it or forcing it into an alliance with the Dual Monarchy—Germany’s options were to support its ally or not, and the likely outcomes all involved some form of Russian involvement. However, contrary to some accounts, it doesn’t look like Germany was bent on provoking a war with Russia; to be sure, its leaders had been wondering whether preventive war was attractive since Russian rearmament began in 1905, but it had always used Russian assertiveness as the indicator of whether the time had come to attack before the Great Program of rearmament would be complete. In 1905, 1908, 1911, and 1913, the indicator never appeared, and preventive war was never attractive. In short, Germany would fight a preventive war when Russian assertiveness indicated that the time was right, but it was hardly bent on it from the start. As such, we came up with the following ranking, from best to worst:

  1. Austria attacks Serbia, Russia doesn’t intervene (Serbian power eliminated, Dual Monarchy secured, Russia pushed out of the Balkans, and costs of war saved, Entente might fall apart)
  2. Austria attacks Serbia, Russia intervenes, Germany fights (Russian assertiveness reveals preventive war’s time has come)
  3. Austria attacks Serbia, Russia intervenes, Germany doesn’t fight (closest ally defeated, Russian interests ascendant in Balkans, and Russian growth continues unchecked)

Why specify these preferences? Two reasons. First, it lets us see how they “add up” and interact with Russian and French preferences to get us the World War that we had emerging from the July Crisis. (Hint: the revelation of Russian assertiveness turns out to be key in convincing Germany that it’s time to preventive war.)

Second, and more important for today, it’s a useful exercise in thinking hard about how we explain political phenomena. If Germany’s best possible outcome was “localization,” as contemporaries called it, where the threat to support Austria kept the Russians on the sidelines and ensured that the Triple Alliance would dominate the Balkans in the future, then it’s hard to assert—as a few still do—that Germany wanted a preventive war, pursued and then got its preferred outcome. It’s an easy argument to make, but it’s probably also fallacious: Germany seemed to have gotten its second best outcome, and we can’t infer from what did happen that this is what Germany wanted to happen. This means that, to understand why the July Crisis turned into the Great War, we need to know something about why localization failed—why the German threat to support Austria-Hungary was insufficient to deter Russia from mobilizing and continuing the Continent’s slide into war. If, on the other hand, we assume that what happens is exactly what people want to happen, then we can tell a nice linear story—but we also run a serious risk of mis-explaining one of the key events in world history.

Here’s a brief example: after the British indicated that they would have trouble remaining neutral, Chancellor Bethmann Hollweg actually tried to convince Austria to “halt in Belgrade,” limiting its own preventive war to a punitive raid, and while this is consistent with the idea that localization was the goal (British support would certainly embolden the Russians and the French), it’s an anomaly for the idea that Germany was bent on war. Thinking hard about preferences, choice, and strategic behavior, then, can clearly improve our understanding of even the biggest, most complex of events.

Just how German preferences interacted with Russian and French preferences to produce the War, though, will be the subject of the next lecture.

Battle Plans, Strategy, and Equilibrium (WWI in Real Time, Lecture 4)

It’s hard to talk about the First World War without at least a nod towards the war plans and strategies that the belligerents carried into the war—and that subsequently went pretty poorly for nearly all of them. Today, following Tuesday’s discussion of the crises that made a general war feel inevitable, we spent some time in the heads of the generals and civilians that would plot out how such a war would be fought. However, rather than look at France’s embarrassment in the Battle of the Frontiers, Germany’s disaster on the Marne, or Austria’s failure to crush Serbia to judge war plans, we looked at the strategic environment in which the plans were made—which makes even the Schlieffen/Moltke plan look a lot less crazy than folks often think it is.

To illustrate the problem today—that of planning a battle when you’d like to meet an opponent’s attack, while she would like to attack around your defenses—we rely on the Colonel Blotto game, which looks something like this:


The first thing we establish is that, if we’re looking at pure strategies (that is, choosing one with probability one given some action of the other player), there’s no obvious way to play the game: 1 wants to attack the right (left) if 2 defends the left (right), while 2 wants to defend where 1 attacks. As a result, each possible pair of choices has a profitable deviation; one player wants to change her action at any possible equilibrium.

How, then, should players (read: military planners) approach this problem? Technically, we say that players solve the problem by playing mixed strategies—that is, randomizing over their actions at a specific rate—but the essential idea behind it is being unpredictable: by making it difficult to guess where I’ll defend or attack, I can blunt any advantage my opponent would have by knowing the placement of my army, and vice versa. Since one’s opponent isn’t sure of where the attack will come from, she assigns probabilities to each one, which is what gives the “mixing” part of the strategies their substantive bite. This, of course, is why planners try to keep their strategies secret, why states work so hard to uncover one another’s plans, and why successfully learning another side’s battle plan can play such a critical role in the outcome of battles.

So what’s the relevance of the Blotto game for military planning ahead of World War I? I’d say it boils down to a few things:

  • since each side can try to blunt the other’s advantage, no one is going to do as well as possible in expectation
  • in fact, every possible outcome (1 running into 2’s defenses and 1 running successfully around them) occurs with some chance
  • these two facts make it very difficult to judge the wisdom of a military strategy ex post, because being unpredictable, being secretive, etc. opens up a range of possible outcomes
  • the best way to judge a strategy, then, is to look at the strategic environment in which it was formed.

This, I think, is pretty instructive, especially in the complex environment of strategizing before World War I. After walking through the solution to the Blotto game, we took each of the principal belligerents in turn—France, Germany, Austria-Hungary, Russia, and Britain—and assessed the extent to which their plans aimed at (a) blunting potential opponents’ advantages and (b) overcoming opponents’ attempts to blunt their own, and it turned out to be pretty fruitful. When we see battle plans—even the Schlieffen/Moltke plan—as compromises among what are often a lot of bad options…we see them a bit more accurately than we do when judging them only on their outcomes.

Next week, though, we get to see these plans put into action—Germany’s attempted envelopment of Paris, France’s attack into western Germany, the Dual Monarchy’s decision to prioritize Serbia over Russia, and the latter’s attempts to shore up Poland by engaging in both East Prussia and Galicia—as the July Crisis erupts and ultimately turns into the Great War. I, for one, can’t wait.

Europe before the Great War(WWI in Real Time, Lecture 3)

After finishing last week with a discussion of the logic of the Anglo-German Naval Race, we did a more inferentially-minded exercise today, examining the four “big” great power crises of the early 20th Century: First Morocco, Bosnian Annexation, Second Morocco, and the First Balkan War. The reason we talk about them is quite simple: to understand why war happens, we must compare it to instances were war was averted, lest we find ourselves affirming the consequent in some bizarre claim that, say, great power crises (or armies, or uniforms, or maps, or alliances…) cause major war. So today was just as much about the logic of inference as anything else, but we got to do it by building up a narrative about the crises that led up to the war, revealing information over time about the extent of the Russian recovery from defeat in the Russo-Japanese War, both materially and over its willingness to impinge on Austrian interests in the Balkans.

The basic logic went something like this: in each of these crises, we saw great powers willing to match each other’s cooperation (that is, avoiding open conflict) with cooperation, and they were able to signal peaceful intent at key points by limiting their military mobilization—in some cases, foregoing substantial amounts of combat power by releasing conscript classes they could otherwise have held over (a costly signal of peaceful intent, indeed). Information problems arose over one great power or another’s willingness to fight, and they were able to use costly measures to back up statements of intent; interestingly, though, the challenge for the offensively-minded governments of early 20th Century Europe, it wasn’t about credibly signaling resolve but a lack of it—it was more difficult to say, “no, we really won’t fight” than to say “yes, we really will fight,” which might go some way to explaining why costly signals of reassurance, not costly signals of resolve, went a long way to explaining the peaceful resolution of these disputes.

It’s also instructive, I think, with respect to the preventive motives that would ultimately push Germany towards war; from 1905, when Russian weakness made France pessimistic about resolving the First Moroccan Crisis with war, to 1913, with the Russian economy recovering at a good clip and rearmament with French funds proceeding apace, Germany seemed to be watching for one thing: an indicator that Russia was rising fast enough—in arms and willingness to use them—to make a preventive war attractive. By 1914, when Russia tipped its hand over willingness to intervene on Serbia’s behalf, Germany’s information problem, which was a question over whether the time had come for a preventive war, was solved; we moved from a world dominated by informational problems, resolvable by costly signaling, to one dominated by the commitment problem, where uncertainty is beside the point.

Oddly, the July Crisis ultimately revealed information about Russian resolve, but the revelation of information isn’t necessarily a force for peace when what one learns is that preventive war has no become attractive

…but I’m getting ahead of myself. [Slow down, Scott.] Tomorrow, we pick up in 1913-14, as war looks inevitable to some on the Continent and demands that they firm up plans and strategies for how to fight it…