World War I in Real Time: First Edition Problems

I’m teaching out of my World War I game theory textbook for the first time this semester, and as I worked through some early-morning class prep this morning, I noticed a tiny little error in Chapter 2. But worry not: all the equilibria still exist, and for the same reasons. It’s a typo, but one worth clarifying.

What’s the issue?

Section 2.2, “Commitment Problems and War,” motivates a game in which state A has to decide whether to launch a preventive war in light of the possibility that B, who’s rising in strength, might renege on the status quo in the future. And in describing A’s payoffs, I say that its best outcome of a peacefully honored status quo gives it 4, its worst outcome of passing on war only to see the status quo renegotiated is 1, and the middling outcome of launching a preventive war today is 2. That’s true whatever B’s strategy happens to be, because that strategy is preempted by A’s use of war.

That … makes sense.

But for some reason, in Figure 2.5 (rendered below), A gets 2 for the outcome of the (attack; honor) strategy profile and 3 for the (attack; renege) strategy profile. That’s…unnecessary. A should really get 2 for both, unless we want to say that attacking a B that would’ve honored the agreement is regrettable, but that’s not necessary for the story. It’s also not in my initial description of the payoffs.

Screen Shot 2020-01-30 at 5.39.22 AM

Now, as you’ll see, the Nash Equilibrium of the game (marked by the solid gray lines) is the same whether that offending 3 is in there or is replaced by the intended 2:

Screen Shot 2020-01-30 at 5.50.08 AM

But I don’t want to let errors like this pass without some kind of note to adopters and students and whoever else has posts on this blog inflicted on them. So, apologies, dear reader(s). Let’s hope there aren’t too many more posts like this one forthcoming.

My own buffoonery aside, there’s a useful point here: forcing ourselves to “do the math” means we can more easily find, correct, and assess the consequences of mistakes in our premises and/or our reasoning. That’s always and everywhere a good thing for the social scientist.

Why Did Belgium Resist in 1914?

Another entry in a series of blog posts about World War I, political science, and my textbook: The Politics of the First World War: A Course in Game Theory and International Security. You can read the previous entry here.

On 3 August 1914, Belgium refused the previous day’s “request” that Germany be allowed to pass through its territory on the way to a clash with France in the opening days of the First World War.

Germany’s note made the consequences of refusal pretty clear:

Should Belgium oppose the German troops, and in particular should she throw difficulties in the way of their march by a resistance of the fortresses on the Meuse, or by destroying railways, roads, tunnels, or other similar works, Germany will, to her regret, be compelled to consider Belgium as an enemy.

In this event, Germany can undertake no obligations towards Belgium, but the eventual adjustment of the relations between the two States must be left to the decision of arms.

Its request ultimatum denied, Germany duly invaded Belgium on 4 August, and King Albert’s troops mounted a gallant resistance, slowing down Moltke’s right hook as it tried swung around the main French force in an (ultimately vain) attempt to create a modern-day Cannae.

Yet the question of why Belgium resisted near-inevitable defeat rather than accept Germany’s terms, which included promises of restituted territory and financial compensation, looks puzzling after the fact.1 Why did Belgium send its modest army into the field only to suffer entirely predictable death, destruction, and dislocation, to be pushed back to Antwerp as the Germans did what they pleased? And why forfeit the chance that Germany would follow through on its ostensibly generous promises of restitution and compensation?

Reiter and Stam, defending a unilateral model of war onset, put it down to national honor: better to go down fighting than surrendering (p. 380).2 Vasquez, in his wide-ranging and fascinating book on the spread of the war, also nods to national honor but argues further that Belgium’s resistance is an “anomaly” for bargaining theories of war (pp. 111-114)—that is, for theories that view war as a wasteful alternative to negotiations that can (and often do) produce the same outcomes as war-ending negotiations.

It won’t surprise you that I disagree with the “anomaly” take. Here’s why.

The version of bargaining theory Vasquez addresses makes a couple of basic points. First, since war is costly, states have strong incentives to avoid it by striking deals that reflect the likely outcome of a war. Second, if (a) states don’t know what bargains others will accept in lieu of war or (b) bargains struck today will only get worse in the future, then states might knowingly fight costly wars. (For more, I recommend Phil Arena’s venerable breakdown of the seminal piece in this literature.)

Let’s take a look at Vasquez’s argument. He first notes that Germany’s ability to overwhelm Belgian defenses, including the forts at Liège, wasn’t much in doubt (pp. 111-114). And, in the absence of shifting power that would make that bargain obsolete in the future, we’d expect Belgium to cut a deal; why fight a costly war when the offer on the table reflects the likely outcome minus all the death and destruction? What’s more, Germany’s offer of restitution looked pretty generous; Belgium could hardly hope to achieve that much by fighting. So far, so good. Acceptance would be consistent with bargaining models of war…

…but only if Belgium believed that Germany would’ve honored those promises.

My issue with the anomaly claim is that it takes as credible the German prewar promise to evacuate Belgian territory and provide compensation after the war. But unlike the actors in the drama he recounts, Vasquez treats German talk credulously. He argues that Belgium should’ve expected Germany to honor its pledge, citing as evidence one of Chancellor Bethmann-Hollweg’s post-invasion Reichstag speeches (a) apologizing for the violation of international law and (b) reiterating restitution promises. Vasquez’s argument that Belgium should’ve accepted Germany’s terms rests on Germany’s 1914 description of its own war aims…precisely when it had incentives to lie about just how expansive its aims were.

Yet one of, if not the, most important parts of the story in August 1914 is that those on the receiving end of German promises about limited aims didn’t believe them. I argue in The Politics of the First World War that one of the animating strategic problems on the soon-to-be Western Front was the question of whether Germany was, as it claimed, fighting a defensive war or, as its actions indicated, using the opportunity of a general war to unhinge the balance of power (see Chapters 4-6 in particular).3 The problem for Britain and Belgium was that Germany would’ve made these same promises whether or not its aims were truly expansionist. Whether we might be able to say after the fact that Germany’s aims were limited (and I’m not sure we can, tbh), what we see with hindsight doesn’t necessarily offer clues about what the actors in our story were thinking in real time.

Incentives to lie create incentives to disbelieve. And Germany’s opponents did just that: they disbelieved.

Yet Vasquez asserts that Belgium should’ve believed Germany in 1914. And this argument that Germany wouldn’t “go back on its pledge” to restore Belgium hangs on a claim that power wasn’t shifting between Germany and Belgium (p. 112).

As noted above, this requires us to credit Germany’s words when Belgium, to say nothing of Britain and France, didn’t. The question for those watching (and fearing) the approaching German right hook was what Germany would do with a prostrate Belgium if given the chance. And protestations of defensive intent simply weren’t credible as German footfalls echoed through Liège, Namur, and Mons.

The claim that power wasn’t shifting between Germany and Belgium elides the fact that, once in control of some chunk of Belgian territory, Germany would’ve been in prime military position to refuse its return. Merely by holding Belgian territory, Germany’s power over Belgium would’ve increased substantially. “Friendly” language in ultimatums, as Vasquez has it, and ex post apologies (as Bethmann-Hollweg hoped to have it) aren’t evidence that the Belgians believed other than they did, i.e. that a Germany aiming to reorder great power politics wouldn’t have relinquished its gains. (You show me an apologetic speech in the Reichstag, I show you the Septemberprogramm.) If the Kaiser held Belgium, Belgians and French and Britons believed, he’d have every reason to hang onto it. And whether we can say that Britain, France, and Belgium were wrong over a century after the fact doesn’t matter if we want to explain Belgian resistance in 1914; it matters what Belgium and France and Britain believed in 1914.

And Belgian resistance responded to a clear commitment problem: whatever promises Germany made in early August would’ve come due after German troops occupied Belgium. Attractive those terms may have been, but credible they weren’t. Demanding tomorrow the withdrawal of a German army already sitting on Belgian territory looked like a fool’s errand today, making war (with eventual Entente support) more attractive, even with long odds in the short term.

Unilateral control of Belgium would’ve allowed France, Germany, or Britain to menace the others, and that that fact wasn’t lost on France, Germany, Britain, or Belgium.4 And as I argue in Chapter 6 of “The Politics of the First World War” and a forthcoming article at Security Studies, violations of 1839’s Treaty of London guaranteeing Belgian neutrality indicated to observers that the violator had aggressive ambitions—i.e., a desire to overturn the European balance of power—even in the face of likely balancing responses by the other powers.5

So, if we take into account

  1. the fact that occupying Belgian territory would’ve shifted power dramatically in Germany’s favor, making it very easy to refuse to leave; and
  2. the international-legal environment, which framed the German invasion of Belgium against the 1839 Treaty of London,

we can see that Belgian defiance in the face of Germany’s note isn’t an anomaly for bargaining theories of war. An account of why Belgium resisted follows directly from those theories. We just have to remember that Germany’s contemporaries were skeptical of its promises, whatever we might believe today about the credibility of those promises.

If Belgium resisted to preserve the national honor (and would any belligerent government have said otherwise?), it also had good reason to believe that resistance promised the best (if long) odds on preserving independence. Had Belgium welcomed German troops, it would’ve compromised its independence if Germany won the war; and it might’ve fare less well at the eventual peace conference had Germany’s enemies won the war. In the final accounting, a commitment problem explains both Belgium’s resistance and its skepticism of German promises.

Vasquez’s account, on the other hand, requires us to (a) ignore what observers in 1914 saw and believed about German aims and (b) deny that it’s easier to hold on to most of a country’s territory than it is to take it in the first place. I’m unprepared to do either. Plenty of offers, even ultimatums, can look attractive, but they’re only acceptable if commitments to follow through are believed to be credible. And when the pain of a future weakened bargaining position is greater than the costs of war today, states may take the plunge into costly war…just as Belgium did in 1914.

TL;DR Belgian resistance to Germany’s 1914 ultimatum isn’t an anomaly for bargaining theories of war. States sometimes have good reason to fight in the face of unfavorable odds if the peaceful alternative is near-certain disaster, and King Albert’s decision is a compelling case in point.

  1. To call back to an earlier post, hindsight is great for generating puzzles but often shit for solving them. ↩︎
  2. I don’t think they needed to make that argument, though. War occurs because the side that wants war chooses to over-demand, not because the victim views defending “one’s own land as an insubvertible good” (p. 380). Countries can still have wars when they want them; they’ve just got to demand (or try to take) more than the other side will yield. ↩︎
  3. And not just in Europe, but globally. This was a bid for world power, in the sense of competing with other global empires, as Fischer notes; but it was a bid to dominate the pre-World War I world of empires, not the one produced by World War I. ↩︎
  4. Belgium was a valuable buffer state, a source of bargaining power to whoever controlled it. ↩︎
  5. Here’s an older post on the topic from the first time I taught “World War I in Real Time.” ↩︎

Why a Textbook on World War I?

Over the next few months, I’ll be blogging about my just-released textbook, The Politics of the First World War, as well as what I learned about international relations by putting the course together, teaching it, and writing the textbook that collects its lectures. (You can find these posts by clicking on the #textbook tag.)

There’s a not-unpopular sense in which studying the First World War isn’t just passé but possibly even misleading, at least for those of us that like to make general statements about the workings of international politics. It’s been done to death, used to generate too many conjectures and theories. Nor can such an outrageous outlier be all that representative of everyday politics of (and conflicts in) the international system. All told, the Great War is an example of charismatic megafauna, the giant panda or the tiger all the kids want to see at the zoo but that doesn’t represent the rest of the animal kingdom all that well.

And yet I taught a course, then wrote a textbook, about The First World War. What was I thinking?1

It’s worth answering that question, because the outlier claim (which I think I saw William Spaniel make somewhere among the internets) and Cullen’s giant panda argument have legs if we’re trying to use the Great War as a jumping-off point for half-baked generalizations.

But that’s (thankfully) not what I’m doing here.

First, the textbook uses the war as a device to teach about international security and game theory in general. Science is very often about the identification of non-obvious commonalities between ostensibly different things (as opposed to, say, the exhaustive cataloging of fairly obvious differences). And the textbook turns the collected wisdom of modern theories of war and politics back on the First World War. It’s an example of many strategic tensions and fundamental political concepts, but it’s not even the only one. I also spend time on the Korean, Russo-Japanese, and Chinese Civil Wars, the politics of nuclear weapons (it’s my book, so I can turn on the time machine when I feel like it), etc. But the goal is to show that things that seem unique, like the July Crisis, which Christopher Clark rightly calls perhaps “the most complex [event] of modern times, perhaps of any time so far,” are really special cases of things far more general…things about which political science already has some good ideas.

Second, the book is ruthlessly committed to a “real time” approach to explanation, casting off inherited hindsight for a laser-like focus on the incentives and constraints and uncertainties that confronted the characters in our story as they faced them. Generals and officers settling into (and stuck in) strategies of attrition. Chancellors and Kaisers, Tsars and prime ministers, debating the merits of war and the credibility of bargains on the table. Neutrals weighing belligerence as the war crept ever closer to their territories. Coalition partners wrangling over burden-sharing and who would be dominant in the postwar negotiations. Titanic strategic gambles, like unrestricted submarine warfare and the Kaiserschlacht, that promised only victory or defeat. Victors redrawing the map of a damaged world, balancing the needs of their own recovery against the hope (vain, as it turned out) to prevent such a thing from happening again. They all look different, and radically so—often more sympathetic, sometimes less—once we shed a century’s worth of hindsight and efforts by the players after the fact to shape the narrative. Approaching and explaining the war this way was a wild ride, and I learned as much as (if not more than) my students in crafting the course, giving the lectures, solving the games, and writing the book.

So, to those who rightly worry that yet another case study of World War I might invade your IR classroom, rest easy. It’s about what we can learn about the First World War by using the collected wisdom and tools of modern theories of war, and it recognizes—explicitly and often pretty aggressively—this basic truth of social science:

Hindsight is excellent for generating puzzles, but it’s very often shit for solving them.

Sadly, no giant pandas here.2 But there are plenty of indulgent Star Wars footnotes, incisive Rambo references, a gloriously apt Eddie Vedder epigraph, and a bonus story about Stalin murdering his pet bird with a pipe. Enjoy.

  1. You mean apart from “the students will totally be into this at the centennial, and maybe I can trick them into doing math”? Sure. I can talk about that. ↩︎
  2. But give me birds any day. They’re dinosaurs! ↩︎

The Politics of the First World War in the Classroom

The Politics of the First World War in the Classroom #WWIinrealtime

#wwi #textbook

My textbook, The Politics of the First World War: A Course in Game Theory and International Security, is out this week (Amazon). And I figure it’s worth talking about how I envision it being used in the classroom. The most important thing to note is that the textbook is supposed to be a bit, in Cambridge’s words, “weird.’’ It’s designed to do things that other IR textbooks don’t do, putting deep engagement with a historical case alongside a survey of the literature on international conflict and an undergraduate game theory course.

That ain’t normal.

But the book’s uniqueness puts the onus on me to convince instructors that it need not be used only for something like my “World War I in Real Time” course that inspired it. As a representation of said course, we can generously call this book “abridged,” apart from the game theory content. The relatively limited World War I content is more feature than bug, though, because those class sessions for which instructors don’t assign a chapter (it’s got only 15, after all) can go one of two different ways, creating two rather different courses:

  1. A game theory course. Deeper dives into game theory, even if only to get more practice with what’s already there or to expand the set of applications. (Practice is everything when it comes to writing and solving your own models. Everything.)
  2. An international security course. Assigning some of the journal articles and books that the chapters reference in their penultimate sections (or stuff in the same area), which the students can then use their developing theoretical skills to evaluate.

I’ve taught this course both ways over the years, and each has its merits. In this post, I’ll share some of the strategies I’ve adopted in both versions of the course, including two sample assignments, to give instructors a sense of what worked for me as I trialed and erred my way through early iterations of this material.

A game theory course

Choosing the right level of formality for the book’s game theory component wasn’t easy (and I owe Toby Rider a debt of gratitude for helping me make some crucial decisions on this, especially in the war termination chapters). Modelers will notice pretty quickly some things left out or treated only superficially, like iterated games with punishment strategies or restrictions on out-of-equilibrium beliefs in Perfect Bayesian Equilibrium.

Off-textbook sessions can easily dive into these topics, though, because the main-text material develops the foundations of equilibrium reasoning (the big payoff, IMHO) and emphasizes basic math practice. Chapter 11’s treatment of the Christmas Truce, for example, discusses how to use punishment strategies to induce cooperation in the Iterated Prisoner’s Dilemma (a workhorse model in the study of cooperation). But I don’t solve that game, leaving that for further skills development if the instructor wants to make this a game theory course. It’s easy to imagine an off-textbook session on how different punishment strategies (tit-for-tat, penance, grim trigger, etc.) induce cooperation of different quality and variable robustness under different conditions, allowing students to get extra practice with Subgame Perfect Equilibrium and to think about additional applications of repeated games.

An international security course

Games may be the centerpiece of each chapter’s theoretical work, but they need not dominate off-textbook sessions. If instructors prefer to highlight the international security component, the final few sections of each chapter contai reviews of the related literatures, from arms races (Chapter 3) and solutions to the collective action problem (Chapter 7) to the role of national leaders in international politics (Chapter 13) and the relationship between democracy and peace (Chapter 14). I survey the contemporary literature on these and numerous other topics, and instructors can assign a sample of the cited work to give students more direct exposure to the scholarly literature. An instructor might, for example, assign empirical work to expose students to the ins-and-outs of interpreting quantitative models or evaluating case selection in qualitative models. I’ve taken both tacks, and the book’s focus on rigorous theory helps students encounter unfamiliar empirical methods with appropriate standards of judgment—not just the all-too-pervasive (and unjustified) “math”-phobia.

Another option for this version of the course is to assign supplementary historical readings as applications of ideas introduced in the text. The book’s discussions of coordinating limits in the Korean War (Chapter 10), the termination of the Russo-Japanese War (Chapter 12), and the duration of the Chinese Civil War (Chapter 11) are all good candidates. But instructors may also assign histories of whatever crises, wars, and near-wars they know well—from the Cuban Missile Crisis to the War of the Pacific to the Iraq War—and encourage their students to analyze them in light of the theoretical models and literatures they’ve encountered in the textbook. I’ve not taught this analytical-history version of the course, but I plan to the next time I get a chance—sadly, in spring 2020.

Some assignments

Finally, I’d like to highlight two assignments that I’ve trotted out in both versions of the course, each of which the students (and this instructor) take to pretty well.

First, as an application of equilibrium reasoning and how it helps us think through counterfactuals, I like to give students the following writing prompt after Chapter 8:

Put yourself in [Chief of the German General Staff] Moltke’s shoes and come up with an alternative strategy—either in the initial execution of the war plan or in managing the offensive once [French General] Joffre orders the Allied retreat—and then analyze whether your alternative might’ve made a substantial difference in the outcome of the invasion of France in 1914.

Armed with some practice using logical tools and equilibrium reasoning, students typically don’t use this to engage in the loose speculation that every professor rightly fears would happen in response to such a question. (We can’t all be Harry Turtledove, who seems to just come by this stuff naturally.) Rather, the course’s focal formalities prove to be a nice source of discipline for counterfactual, “what-if?” reasoning. Students have to think hard about why countries attack who they attack and how and when they attack them, how likely targets respond, how that influences initial attack decisions, etc. Seriously, this might be my favorite thing I’ve ever assigned: the answers, as you might imagine, are often pretty varied (some workable, some not, and very many of them delightfully creative). But when we discuss and evaluate their alternative strategies as a class, I’m always impressed at my students’ ability to surprise me; with a shared set of logical standards, their discussions in the post-assignment session are insightful, productive, and just fun.

Second, I like to do a puzzle-focused class session, where students submit beforehand their own historical puzzles based on the First World War (this when they’ve been reading narrative histories of the war throughout the semester—I’ll have another post on my preferred sources for that version of the course soon). I sort (and edit for clarity where necessary) the submitted puzzles, then lead a discussion in which we devise collectively models to resolve those puzzles. We specify and solve these proposed games as a group, and it’s a blast. Students feel some legitimate ownership of the course material, and they see the fruits of using the theoretical tools they’ve been learning and practicing through the semester. It’s a chance to see their developing skills at work, of practicing the procedures of model-building that the chapters themselves cover at length. Lightbulbs go off in that session, and as an instructor, it’s superfluously fun to watch. I can’t recommend this exercise enough…but only in the latter part of the semester. Too early, and a few students will get lost; I like it as a kind of pre-Armistice exercise.

I’ll have more to say about the First World War, as well as the textbook, how I used it, and how other instructors might in the coming weeks and months. But I hope this initial post gives a flavor of just how flexible the book can be, its admitted weirdness notwithstanding.