Symbolic victories and strategic risk

Abstract

This article studies the rational side of symbolic victories. It opens with the broad question, why are some battles more significant than others? Extending the literature on bargaining and war, it argues that a belligerent can deliberately increase strategic risk in order to communicate its strength. By increasing the information a battle conveys, the belligerent artificially creates the conditions for a symbolic victory. In short, strategic risk becomes a useful, costly signal. This claim is developed in a formal model in which players choose between more and less dangerous military options. Under most conditions, a symbolic equilibrium exists in which both strong- and weak-type players are able to signal their types after only one round. This equilibrium’s rapid information flow is unusual in the rationalist literature: typically, strong types must wait to signal effectively. The article goes on to establish that, when the prior probability a player is strong is sufficiently small, this symbolic equilibrium uniquely satisfies the intuitive criterion. It then applies the model to two famous episodes from military history, the Doolittle Raid of WWII and the battles of Cannae and Capua of the Second Punic War. For both, it highlights how actors deliberately manipulate strategic risk to communicate with adversaries, allies, and their own publics.

Keywords

bargaining conflict military strategy signaling

Introduction

What makes a battle significant? Clearly, some battles are more important than others. Criteria to distinguish important battles from unimportant ones, though, are not so obvious.¹ This article expands the literature on bargaining and war to offer one possible criterion: rational symbolism. A symbolic victory becomes more important than other battles by communicating more information. Aware of this potential, a leader can artificially create the conditions for a symbolic victory by deliberately increasing strategic risk. A bold attack against a dangerous objective can separate a strong actor who is willing to accept the risk from a weak actor who is not. Doing so can facilitate negotiation by signaling to an enemy; it can also improve morale and shore up alliances by signaling to friendly and neutral parties. In short, strategists can calibrate military risk to send political messages to allies, enemies, and their own citizens.

Many engagements derive their significance from the information they communicate, not their military impact. In US history, the best example is perhaps Washington crossing the Delaware: while militarily the subsequent victories were minor, they signaled that Washington’s army was stronger than it had seemed before. Washington wished to demonstrate American capabilities to a worried and (justifiably) skeptical public, and he succeeded. Later, Saratoga would play a similar role, giving the French sufficient confidence in American capabilities to join the Revolutionary War. In the case of Saratoga, this symbolism, while immediately recognized by both the Continentals and the French, was probably not the primary reason for fighting the battle; the engagement followed as much from military as political imperatives. Yet, because battles can serve as effective signals, political leaders have an incentive to choose their battles to capitalize on their potential information. Leaders can deliberately seek what, elsewhere, happens by chance. They can deliberately seek to create symbolic victories.

To study the rational side of symbolic victories, I turn to bargaining models of war. Like Clausewitz, bargaining models begin with war’s political dimension, and so they treat battles as stages in a process of costly negotiation.² As an instrument of rational, political bargaining, war can have two explanations: either the belligerents disagree about their relative strength/resolve (an information problem) or one belligerent cannot credibly promise to honor an agreement in the future (a commitment problem).³ Because it can address these problems, war, though a last resort, can still offer an effective tool for political bargaining.⁴

Following this reasoning, we might conjecture that important battles are ones that do more than others to address these underlying political problems. Bargaining models, though, have paid relatively little attention to battles themselves. Especially among formal modelers, the tendency to simplify the choice of battlefields has led to the treatment of battles as equally significant (or insignificant). To begin addressing this gap, I present a formal model in which military leaders are able to choose their battlefields. This choice helps recover some of the symbolism inherent in leaders’ wartime decisions. These decisions allow battles to communicate substantially different amounts of information, showing how some battles might do more than others to solve the informational origins of war.

To illustrate the model, I study two famous encounters, one from the modern world, one from the ancient: the Doolittle Raid and the Second Punic War. In both cases, strategic planners deliberately pursue unnecessarily dangerous engagements to alter the rational expectations of key political actors. Both cases involve militarily inconsequential battles that, nevertheless, greatly affected actors’ strategic calculations. I use the mismatch between military and political value in these cases to isolate a more general phenomenon: battles involving substantial danger often communicate substantially more information. Many strategists, aware of this fact, boldly (or recklessly) increase strategic risk in order to enhance the information their strategies reveal.

Of course, part of a battle’s symbolism is non-rational.⁵ This non-rationality, though, should not lead us to neglect its rational dimension. Concepts like morale and confidence are psychological in nature, but they also reflect rational expectations. In his work on focal points, Thomas Schelling theorized that competitive equilibria would tend to rest on ideas with imaginative salience – that rational optimization would center on fundamentally aesthetic concepts (Schelling, 1963: 57–58). In that spirit, this article deals with the rational side of symbolism, not to understate or dismiss psychology, but to show how the rational and non-rational might be more intimately related than often recognized.

Game theory offers an excellent and underused tool to conceptualize battles as a means of communication. Within the formal literature, an established and flourishing tradition explores signaling and crisis bargaining. Nonetheless, this literature focuses almost exclusively on signals sent by domestic politics (like audience costs) or negotiating strategies (like screening offers). Formal models rarely examine how military strategies themselves might signal strength. By and large, formal models allow a battle’s informational content to have only two sources: the information revealed from the fact that a state won (or lost) a battle, and the information revealed from its willingness to fight. The actual battlefields and the character of the engagement are taken as given: states choose whether to fight a battle, but not which battles to fight. This simplifying assumption has led theorists to neglect a key source of information during war: the choice of where, when, and how to fight. This article argues that, by studying such decisions, we can understand how some victories become symbolic. Before turning to this argument, though, I first want to suggest two empirical shortcomings with existing work and how a model of symbolic victory might address them.

Most formal treatments of information and war assume two actors with common priors and private information about their relative strength; generally, this private information is asymmetric. These models commonly share two features: (i) a screening process leads weak states to accept bargains early in war and strong states to hold out for better settlements; and (ii) information flows primarily through bargaining (i.e. the offers states make and accept), not through battles themselves.⁶ We can draw two conclusions from these features.

First, according to this approach, wars should last longer when an opponent is much stronger than expected, because such an opponent will wait to negotiate sincerely. This prediction is difficult to assess empirically. Without a clear definition of strength, we risk post hoc rationalizations in which any losing state is presumed weaker than its adversary (and weaker than it imagined itself ex ante). On the other hand, when studies of asymmetric warfare find that weak states win wars more often than strong ones, this seems equally unsatisfying – clearly our use of words like ‘strong’ and ‘weak’ lacks clarity (Arreguín-Toft, 2001: 97). Nonetheless, even without a rigorous test we should find this prediction implausible. In sports and daily life, the larger the gap between expectations and reality, the faster that gap becomes apparent (usually). The same seems true among nations. If war begins because of a misperception about power or resolve, then the wider the actual gap, the more quickly leaders apprehend it – as in the Gulf War or the Franco-Prussian War.

Second, because information primarily flows from states’ willingness to continue fighting, according to current models the likelihood of war termination should improve with war’s length. This prediction is more testable, and it seems to lack empirical support. For instance, Ramsay (2008) tests a pair of related hypotheses: (i) that as battlefield information (measured in battle-days) increases, so does the probability of war termination; and (ii) that more shocking victories (measured by the difference of expected and actual losses) increase the probability of war termination. He finds no support for either hypothesis:⁷ shocking victories have some effect, but the variable’s magnitude remains statistically insignificant, while battlefield information, though significant, in fact decreases the probability that war ends.⁸

The argument in this article sidesteps or resolves these empirical problems. First, it shows why wars might be brief when expectations diverge widely from reality: the greater the divergence, the more incentive to fight highly informative battles. Second, it relocates battlefield information away from length of days or casualty figures, instead emphasizing the political calculus behind strategic decisions. By doing so, it suggests how battles communicating substantial military information, such as unexpectedly large defeats, might be less informative (less symbolic) than battles chosen because of their risk or political significance – because war is primarily about politics, not violence. In short, the argument tries to return our focus to the interplay of politics and military strategy.

Argument and model

I argue an actor can voluntarily diminish its strategic odds of victory in order to signal its strength. It enjoys this opportunity because it can afford more strategic risk than a weak actor, and so it can accept dangers that a weak state would not. As a consequence, either its friends or its enemies (or both) can infer from its strategy that the state is powerful. In short, strategic risk can create the opportunity for a symbolic battle; this symbolic battle conveys substantially more information than a standard engagement.

This argument builds on three premises: that leaders may sometimes choose their battles; that they are uncertain about others’ strength; and that war is costly. The last are standbys in the game theoretic literature on war, but the first is somewhat novel. Formal models tend to treat war as a line, as a set of rails from which the belligerents cannot deviate. In these models, generals have no choice over their battlefields or objectives.⁹ In effect, these models eliminate military strategy. By introducing one element of military strategy – the choice of how many hazards to accept on a campaign – I show one way military strategy can communicate political information.

This point deserves elaboration. Many bargaining models of war assume a line of N forts (or objectives); when a player captures all the forts, he wins, seizing the pie in full (Smith, 1998; Slantchev, 2003; Smith & Stam, 2004). In each round, players bargain and, if that fails, they fight for the next fort (or, in the most sophisticated models, they do nothing). These models then find, counter-intuitively, that one’s negotiating strategy conveys more information than who wins on the battlefield. However, this finding is not as surprising as it seems. Since all types of player must attack the same target, information only updates through victory, which is highly variable, and negotiating strategies, which are less so. Thus, in this approach, separation between weak and strong types of players will occur mostly in the bargaining phase, not through fighting itself. These battles are weak signals because there is no choice: because both weak and strong states must attack the same targets, in effect these models impose pooling strategies on states. If we allow for choice, permitting states to attack one target rather than another – if we reintroduce military strategy – then battlefield choice regains its symbolism.

A strong state, I suggest, will undertake danger unnecessarily in order to demonstrate its strength. In effect, strong states will sometimes risk defeat more to prove their capabilities than to achieve a military objective.

A few circumstances can make symbolic victories especially useful. If current strategy faces diminishing marginal returns in the information it can communicate, then a symbolic victory may be necessary to correct an opponent’s inaccurate self-assessment. (If military strategy is held constant, players’ beliefs should update on average less and less each battle.) Such a victory is even more advantageous to a strategist pressured by an unstable domestic situation. For instance, during the Mexican–American War, President James Polk expected Mexico to acquiesce to its early territorial losses, but Mexico refused to acknowledge the weakness of its position. Polk then faced two options: he could press gradually southward, continuing Zachary Taylor’s successful campaign; or, he could open a second, riskier campaign against Veracruz (including an unprecedented amphibious landing), culminating in the capture of Mexico City.¹⁰ He chose the latter, expecting that it would force the Mexican government to accept the reality of its situation, even if the loss of Mexico City could not in itself decide the war.¹¹

As a general rule, symbolic victories are most useful when an actor’s strength differs substantially from others’ perception of it (see Proposition 2). When facing culturally dissimilar adversaries and target populations, imperial powers employ symbolic battles to highlight the gap in power as quickly and obviously as possible. For instance, the Roman Republic and Empire regularly undertook symbolic demonstrations to convince Germanic tribes of their strength, as these less-developed groups tended to underrate the power of the Roman legions.¹² The most celebrated of these demonstrations is Caesar’s crossing of the Rhine, a risky military venture which was undertaken for largely symbolic reasons: to demonstrate ‘that Roman armies could and would advance across the river’ (Caesar, 1951: IV.16). In the east, a similar logic may also explain the famous siege of Masada.

While the model will focus on a two-player interaction, symbolic victories also help sway third parties, including domestic audiences and potential allies. Improving morale is one form of such suasion (this will be seen in the study of the Doolittle Raid). Another is political competition over a given population, as in revolutions and insurgencies. Because revolutionaries require a certain amount of popular support, early symbolic victories (like Valmy) help build a movement’s credibility. Similarly, because insurgencies often target the domestic audiences of their antagonists, a symbolic battle can be a more useful bargaining device than military gains.

A common interpretation of the Tet Offensive traces its origins to this strategic logic. Although its initial, defensive approach would gradually erode American and South Vietnamese confidence, the North deliberately adopted a riskier, more aggressive strategy to accelerate this collapse (Summers, 1982: 183–184). By directly challenging prevailing US and South Vietnamese narratives, the offensive would facilitate a speedier end to the conflict. While a tactical failure, the Tet Offensive had its intended strategic effect, decisively reshaping US perceptions of the war.

Modern counter-insurgency has learned the same lesson. Kilcullen, correcting a common misapprehension, writes that ‘winning hearts and minds’ actually refers to a kind of rationalist signaling: ‘“Hearts” means persuading people their best interests are served by your success; “minds” means convincing them that you can protect them, and that resisting you is pointless’ (Kilcullen, 2010: 37). As a consequence, he urges counter-insurgent commanders to seek early victories, which will probably be tactically inconsequential, as a ‘form of armed propaganda’ (Kilcullen, 2010: 38). Because insurgents require popular support, using symbolic victories to change perceptions can be more effective than dramatic combat operations.

For the same reasons, leaders who rely on large coalitions or empires can use symbolic victories to keep partners and neutral states in line. To this end, attacking a particularly dangerous target can communicate to third parties more effectively than attacking the most militarily valuable (or bypassing the threat entirely). Alexander the Great seems to have invoked this logic explicitly in his sieges of Tyre and Gaza: ‘Alexander thought, on the contrary, that the more impracticable it was, the more necessary was the capture; for the achievement would strike great terror into his enemies just because it was beyond calculation.’¹³

A simple example

Before turning to the actual model, consider a simplified example with two neighboring countries, Aggressor and Defender. Aggressor is one of three types: Strong, Average, or Weak; Aggressor knows how strong he is, but Defender doesn’t. Further suppose that Strong countries win battles all the time ( $P r = 1$ ), Average win half the time ( $P r = .5$ ), and Weak almost never win ( $P r \approx 0$ ). If Aggressor can win two victories over his enemy, he captures the capital and imposes his will on Defender. But if he loses even one battle, the invasion fails and he must return home.

At the beginning of each round, the Aggressor presents Defender with an ultimatum, promising to invade or continue the war unless his terms are accepted. Defender accepts or rejects this ultimatum; if he accepts, the game ends; if he rejects, the war continues.

In typical bargaining models, the war would go as follows. Before an invasion, the Defender would cut a deal to satisfy a weak Aggressor, but no one else. Then, if the Aggressor invaded, he would cut a deal to satisfy an average Aggressor, but not a strong one. Thus, in the standard model, every battle reveals almost exactly the same amount of information: the initial decision to invade tells the Defender whether his enemy is weak or not; the decision to keep fighting tells the Defender whether his enemy is mediocre or not. Each battle weeds out one type of player, while leaving the Defender uncertain over the remaining ones. The strongest Aggressors must fight the longest wars, since there is no way for a Defender to know if he is facing a powerful foe until the war is long underway.

This narrative does not seem to match common intuitions. The largest mismatches between enemies become obvious at the beginning of a war, not its end. If one side is distinctly stronger than expected, conflict has ways of quickly making that inequality obvious. Current models are not well suited to explain this rapid information flow.

Now suppose that, instead of having his battlefields fixed, Aggressor has two possible invasion routes. One route, the normal route, goes over a plain, where Aggressor enjoys his standard odds of victory (1 if strong, ½ if average, etc.). The other route, the risky route, goes over the mountains, where Aggressor’s odds of victory go down: say a strong Aggressor has odds .95 of winning in the mountains (down from 1), but average and weak Aggressors never win in the mountains (odds of 0).

In this case, the war would go differently. As before, the Defender would still cut a deal to satisfy a weak Aggressor prior to an invasion. After an invasion, though, he will cut two separate deals: if Aggressor invaded over the plains, he will cut a deal to satisfy an average foe; if Aggressor invaded over the mountains, he will cut a deal to satisfy a strong enemy. So long as the costs of war are sufficiently high (here, $c > .05$ ), a strong Aggressor will reduce his odds of ultimate victory from 1 to .95, invading over the mountains, in order to avoid the costs of having to fight all the way for the Defender’s capital. Thus, all relevant information is communicated by the first battle, the decision of where (and whether) to invade. A victory in the mountains becomes a symbolic battle, since it resolves all uncertainty over players’ relative strength.

Model

The following game tracks standard bargaining models of war with asymmetric information over capabilities, except that a state chooses whether to fight more or less dangerous battles. In equilibrium, strong-type players will incur additional strategic risk (reducing their odds of victory) to signal their strength. Weak players will not. In the model, incurring this additional risk has no military value; it is only a signaling device. To focus on the role of information, there are no commitment problems or indivisible goods, and a bargain once reached is binding.

Suppose an aggressive country, A, desires to seize from some other country B a prize of value 1. Player A is either weak or strong, $θ \in {W, S}$ , with a prior Pr( $θ = S) = π \in (0, 1)$ . (Note that weakness contains no information about A’s strength relative to state B: state A could be ‘weak’ and still be more likely to win a battle than B.) Player A knows whether he is weak or strong, but B does not.

The aggressor A may choose to invade his enemy along any number of routes, $r \in ℝ^{+}$ . This continuum represents the variety of military objectives an actor might target as it advances toward a decisive battle. (The Online appendix offers an example of symbolic victory with discrete invasion routes.) All are assumed to have equal military value, but some are more difficult to seize than others. For convenience, assume that these routes are arranged in order of danger, so that the route $r = 0$ involves the least risk of defeat. Player A’s invasion succeeds with probability $p_{θ} (r)$ , where $p_{θ}$ is continuous and monotonically decreasing (reflecting the increasing danger each target poses), and p_W(r) = 0 for some r. Proposition 1 will show that, even though these targets have the same military value, a strong-type player will not attack the most vulnerable. Instead, he will attack a more dangerous target to differentiate himself from weak-type players.

When confronting a dangerous opponent, a strong actor’s odds will tend not to drop as sharply as a weak actor’s. Colloquially, we might say that strong players can more easily afford a bold or reckless strategy. The reasons for this resilience are manifold; consider two. Because military power exhibits economies of scale, the difference between a strong- and weak-type player will tend to grow more pronounced as an enemy becomes more dangerous (as long as the enemy is not too dangerous). Also, stronger militaries tend to operate effectively in a wider variety of circumstances, preserving their advantages more easily across different domains, theaters, and terrains (like the adaptability of the US Marine Corps). The model captures this idea by requiring that, when invading at the same point r, a strong player’s odds of victory must diminish (weakly) less rapidly than a weak player’s. Formally, for any $r_{1} < r_{2}$ where $p_{W} (r_{2}) > 0$ , then $p_{S} (r_{2}) - p_{W} (r_{2}) \geq p_{S} (r_{1}) - p_{W} (r_{1})$ . For example, the simplest p satisfying these conditions would be a linear function which decreases players’ best odds of victory $p_{θ} (0)$ by r, for example $p_{S} (r) = max {p_{S} (0) - r,0}$ .

The aggressor A must win exactly two battles in order to win a war: first, A must successfully invade B; second, A must capture B’s capital. If A fails in either battle he loses the war, gaining nothing, but if he wins both battles he gains the entire prize. Between battles A presents an ultimatum which B may accept or reject. A battle costs each player some $c > 0$ . (Bargaining takes place after the invasion merely to simplify the game’s setup; in a sense, the game begins after the weakest types have already made their bargains and been weeded out. The model could allow for bargaining before an initial invasion, but it would need to allow at least three types of player, as in the simple example in the previous section.)¹⁴

So the game proceeds as follows. First, A chooses where to invade B, using an invasion route $r \in ℝ^{+}$ . Players fight this battle. If A loses, the game ends, and their payoffs are $- c$ and $1 - c$ . If A wins, he proposes an ultimatum division of the good, $x_{A} \in [0, 1]$ and $x_{B} = 1 - x_{A}$ , which B may accept or reject. If B accepts, the game ends, and their payoffs are $x_{A} - c$ and $x_{B} - c$ . If B rejects, A attacks B’s capital, and the game ends. The winner receives $1 - 2 c$ and the loser $- 2 c$ ; for simplicity, assume A wins this battle with odds $p_{θ} (0)$ .

The model involves a minimum of moving parts. Its key elements are the variable strength of the aggressor, the non-zero cost of war, the defender’s uncertainty about his enemy’s capabilities, and the aggressor’s ability to choose more or less dangerous battlefields.

In order to demonstrate the intuition that strategic risk can communicate an army’s strength, I have assigned to such risk only severe consequences: in the model, each decrease in the probability of a battlefield victory likewise risks the entire campaign. Moreover, no one can mistake the decision to fight a riskier battle as an attempt to conserve resources or incur fewer casualties, since none of these factors are at play. There exists only one trade-off: a strong state must weigh the decreased probability of ultimate victory against the increased returns from signaling its strength. I argue that above a quite minimal cost-of-war threshold, a symbolic equilibrium exists in which a strong state always stands to gain by undertaking unnecessary risks.

Proposition 1 describes an equilibrium in which a strong aggressor can invade at some $r > 0$ , attacking an unnecessarily dangerous target, to successfully signal his type. This equilibrium exists under very mild conditions on the parameters: the costs of war c must be large enough, relative to the gap in strength between W and S, to make a negotiated settlement worthwhile. Proposition 2 suggests that this separating equilibrium is ‘more likely’ than other equilibria because it uniquely satisfies the intuitive criterion.

Proposition 1: Call a symbolic equilibrium one in which different types of players choose different invasion routes, r, and thus reveal their types. A symbolic equilibrium is guaranteed to exist if $c \in (\frac{p_{W} (0) [p_{S} (0) - p_{W} (0)]}{2 p_{S} (0)},1 - p_{S} (0))$ .

A few points are worth noting immediately. First, in typical models, when a game has fewer bargaining rounds than player types it may not be possible to fully separate kinds of players and still negotiate a settlement; the possibility of choosing $r > 0$ creates this peaceful outcome. Second, because of the rapid information flow, war always terminates in a negotiated settlement, not a decisive battle. Third, the threshold requirement for c relative to p_S and p_W is quite lax, never exceeding ⅛, and usually approaching 0; in other words, most real-world circumstances will have a symbolic equilibrium.

A symbolic equilibrium is not unique. There are good reasons, though, to believe it matters. Proposition 2 establishes that, when an opponent does not ‘expect’ an aggressor to be strong, then a symbolic equilibrium uniquely satisfies the intuitive criterion (IC). This result accords with common sense: the less an enemy thinks his opponent is strong, the more reason the opponent has to demonstrate it. The IC allows us to weed out equilibria which rest on implausible off-the-path beliefs. With some abuse of the concept, we can sketch the IC’s role in Proposition 2 like so: if player B observes A attacking a risky target, it cannot assume that the behavior is simply a weak player behaving irrationally; rather, it must ask whether a strong player (and only a strong player) could profit from attacking such a target, and if so, B must update its beliefs accordingly.

Proposition 2: When the prior $π$ is sufficiently small, $π < \frac{p_{S} (0) - p_{W} (0)}{2 p_{S} (0) - p_{W} (0)}$ , then the symbolic equilibrium is the only PBE satisfying the intuitive criterion (IC).

Importantly, in this model the IC suggests that a victory can become symbolic even without a pre-existing framework or convention in which its symbolism would have a clear meaning. Simply put, a battle can become symbolic even if, beforehand, an adversary had little idea or intellectual frame in which to interpret its meaning. A battle’s symbolism does not require a pre-existing interpretive structure to make it intelligible; rational actors will understand its symbolism regardless whether or not they had imagined it beforehand.

This implication highlights a key place where a rational account of symbols will diverge from a constructivist. If we take the IC seriously, then actors can communicate using symbols with dissimilar rivals, because actors can come to understand what a victory symbolizes even if, beforehand, they had no conception it could happen. Symbolic victories will be possible (and sought) by adversaries who lack a shared culture in which to understand them. The rational imperatives of war can create the means of its own interpretation.

Caveats and implications

A few caveats are in order. First, it may be difficult to send as clear a signal as a strategist hopes. In the model, both players observe r accurately and r has no effects other than reducing A’s probability of victory; in fact, the world might not be so transparent. An unnecessarily dangerous battle can signal strength, but an observer might confuse it with a high-risk, high-reward strategy, or perhaps attribute it to stupidity or error. Likewise, the fog of war means that players cannot always observe r, and if a rival cannot recognize risky behavior, such behavior might lose its signaling value.

Second, the model suggests that leaders should sometimes ‘tie their hands behind their backs’. But leaders rarely sink their own ships just to make a point. If the model is correct, the reason for this reluctance is not wholly clear, but we can offer a few conjectures. From the vantage of soldiers and concerned citizens, it would be hard to justify the destruction or non-use of assets that could save lives. By contrast, assaulting an unnecessarily difficult or risky target might be more acceptable, since a leader is not burning resources or recklessly endangering soldiers. Moreover, a leader must be aware that his domestic audience and his armed forces might not be as ready as himself to attribute strategic risk to genius rather than incompetence. He must also be prepared for an opponent who might not understand the signal or negotiate as hoped, and so strategic risk should always entail some reward; that way, if negotiations break down, the strategist has still obtained a valuable and useful prize (as in the Veracruz Campaign). If these conjectures are correct, leaders seeking to create a symbolic victory will use only some of the methods available to them: put simply, they will choose their targets to send a message, but they will not leave their tanks at home. This behavior seems to track that of the previous examples: Alexander’s seven-month siege of Tyre, Polk’s amphibious invasion of Mexico, and North Vietnam’s Tet Offensive were not militarily necessary, but once decided upon they were undertaken with the substantial resources at the strategists’ disposal. To the extent these victories were symbolic, that symbolism derived from the nature of the target, not the means brought to bear against it.

Third, there may exist only one feasible strategy. The argument works by re-introducing strategic choice, and the model allows a player to choose from a continuum $r \in ℝ^{+}$ . In actuality, leaders choose from a finite number of options. As a result, there may no longer exist an r such that a strong state could profitably signal its strength (this is akin to arguments about indivisibility and the onset of war). This caveat suggests why leaders sometimes wait months or years before trying to create the conditions for a symbolic victory: the opportunity may not have arisen.

Finally, the model may apply poorly to absolute wars. If actors expect to fight until unconditional surrender (as modern democracies seem increasingly inclined to do), then a leader has no reason to try to signal his capabilities to an opponent, because he does not expect the war to end in a limited, negotiated settlement. In these circumstances, strategists will simply pursue low-hanging fruit, and symbolic victories, if sought, will be used to signal to third parties (like potential allies or domestic audiences), not to an antagonist.

With these caveats in mind, the model supports at least two related hypotheses. First, attacking unnecessarily dangerous targets should produce informative battles, leading key actors to update their beliefs substantially more than after other engagements. Second, symbolic victories, when achieved, should reduce the length and scope of political bargaining, whether that bargaining takes place between enemies or between allies. Both hypotheses suggest that a symbolic battle will shape a war’s outcome more than other victories, regardless of its scale or duration. In essence, a battle can derive additional significance from its superfluous risk.

I now (briefly) turn to two cases to illustrate the model at work. I use the first case, the Doolittle Raid, to show that the logic I outline explicitly passes through the minds of military strategists. The second case treats the hypotheses in greater depth. Drawing from the Second Punic War, it contrasts the most famous and militarily successful battle of the war (Cannae) with the most politically symbolic (Capua) to show why the latter, not the former, had such an important role in the war’s outcome. While their treatment is brief, I hope that these two cases will suffice to demonstrate that, in ancient and in modern grand strategy, leaders deliberately calibrate risk to achieve symbolic victories, that these symbols are meaningful both to opponents and to allies, and that a battle’s significance can derive as much from its imaginative force as its military achievement.

The Doolittle Raid

A series of Allied setbacks followed the bombing of Pearl Harbor, including the loss of Hong Kong, Singapore, Manila, Guam, and the Dutch East Indies. These setbacks reinforced the intention of President Roosevelt to strike the Japanese mainland. Although such a strike would cause only minimal damage, Roosevelt recognized that it would have an indispensable impact on the American and Japanese psyche. On 18 April 1942, America bombed Tokyo.

The Doolittle Raid is well known, enjoying numerous treatments in popular histories and on the silver screen (all, naturally, of varying accuracy). This fame is unsurprising: its planners intended the raid to excite the American imagination and chill the Japanese; its purpose was not so much to inflict damage as to demonstrate capability. I include it here not to offer an original take on its events but rather because it epitomizes the argument I present. As its details are fairly well established, I hope that a survey of the salient facts will demonstrate the present argument without arousing controversy. Most agree that the highest levels of command, including Franklin Roosevelt, called for the raid in order to boost American morale and demoralize the Japanese; similarly, most scholars agree that it achieved both ends. That is, the US Air Force undertook a great risk for comparatively little military reward and for almost no other purpose than to signal its strength.

The raid was the deliberate policy of Washington. It was not a tactical decision: the assault had its genesis and purpose in the grand strategy of the war. By late December Roosevelt had directed his commanders to plan an attack on the Japanese mainland. The President first approached his Russian allies, whom he hoped would bomb Tokyo for him or, when that tack was rebuffed, at least permit him the use of Vladivostok. The Russians declined, but Roosevelt remained undeterred. In January, he once again pressed his chiefs to devise a way to attack the Japanese mainland, unaware that such plans had progressed steadily since his initial decision. Again and again, Roosevelt emphasized the psychological importance of such a victory (Schultz, 1988: 47), both for Americans distressed by a string of early losses and for Japanese who had not yet tasted serious defeat.¹⁵ A similar logic echoes among the raid’s architects, Duncan and Low, who both agreed that American soldiers and citizens needed a dramatic victory to improve their morale (Schultz, 1988: 14). So their aims were modest, yet highly significant; they were not trying to force the Japanese to negotiate; they only sought to send a message via a symbolic, psychological victory.¹⁶ The United States needed to demonstrate its strength, both to its allies and to its enemies, and no better way existed than a risky, flagrant attack on a homeland thought untouchable.¹⁷

It fulfilled these intentions. Domestically, the attack lifted ebbing American spirits and captured the popular imagination (hence the movie adaptations); internationally, it shattered the confidence of the Japanese people.¹⁸ It had two primary effects on Japanese strategy. First, it led the Japanese to allocate planes for homeland defense, planes which were desperately needed elsewhere, chiefly in the Solomon Islands (Merrill, 1964: 180). Second, it convinced Japanese leaders of the need to extend their defenses. As many have observed, before the raid Japan focused its efforts to the south; after, they turned east, determined to deny the United States the wherewithal to attack Japan again. The result of these strategic blunders was Midway, the turning point of the Pacific War.¹⁹

The Doolittle Raid altered the war – and yet, it inflicted almost no damage. Indeed, upon landing Doolittle himself thought that the attack had failed.²⁰ The tactical futility of the raid has perhaps no better witness than the absence of any intention to repeat it: if its planners had thought it a viable way to inflict material damage, they would have continued bombing; instead, though pleased with its results, no one, including Roosevelt, showed any inclination to replicate the feat (Glines, 1988). Once the raid demonstrated American strength, repeating such a show would serve no purpose. The victory was symbolic.

This symbolism, while meant to communicate with Roosevelt’s Japanese opponents, was aimed even more at the antagonists’ domestic audiences. If morale includes rational expectations of victory, then an easy way to improve morale (or undermine an enemy’s) is to change those expectations. A particularly risky battle, especially a battle only a strong state could undertake and win, can quickly and valuably correct the overconfidence of an opponent or the underconfidence of one’s own people. The Doolittle Raid did both.

The Second Punic War

The Doolittle Raid shows Roosevelt using a risky military venture to signal strength, and in particular to signal strength to the Japanese and American civilian populations. The next case will examine symbolic victories as a tool to signal to allied and neutral states.

In 216 BCE Hannibal Barca achieved the greatest tactical encirclement in the history of warfare. His genius annihilated a Roman force almost twice the size of his own, a defeat that would haunt the Roman psyche for generations. To overstate Hannibal’s tactical achievement would be almost impossible: facing over 80,000 Roman soldiers with perhaps 40,000 of his own, Hannibal set a trap which would kill over 48,000 of his enemy, more than fought in his entire army.²¹ At Cannae, Carthage inflicted the rising Roman Republic’s bloodiest loss.

In this study, I contrast the battle of Cannae with the siege of Capua. I argue that, although the former exceeds the latter in every measure of tactical military success, the latter, not the former, was more significant to the Second Punic War, because Capua, not Cannae, was a symbolic victory. To advance this argument, I first discuss the antagonists’ strategies; second, I highlight the failure of Cannae to alter the war; third, I discuss the strategic logic for besieging Capua; and finally, I contrast the strategic significance of Capua’s fall with the strategic unimportance of Cannae.

The grand strategies of Rome and Carthage are well documented. Roman strategy passed through three main phases in the Italian theater. First, under Quintus Fabius (appointed dictator), Rome sought to avoid open battle while denying Hannibal the resources necessary to sustain his army in a foreign country. In its second phase, Rome abandoned this strategy to engage the Carthaginians directly, leading to its defeat at Cannae. In its last phase, Roman strategy returned to a Fabian approach, eschewing open confrontation while attempting to confine the damage Hannibal could inflict. Rome recognized it did not need to beat Hannibal to win: so long as Rome maintained control over the larger peninsula, Hannibal could win every battle and yet lose the war.

Carthaginian strategy was equally straightforward. Via a ruthless military campaign, Hannibal intended to induce Rome’s allies to defect. Hannibal recognized that Rome drew its manpower from two sources: the city itself, and Rome’s Italian alliances and possessions. Rome’s defenses meant the city would be difficult to take without a protracted siege, for which Hannibal was unprepared; more importantly, were Hannibal to invest Rome, he would lose the key to his tactical victories, his freedom of maneuver, without which he might not defeat the larger armies Rome could raise. Therefore, Hannibal recognized that his only route to victory was to unravel the Roman alliance system.²² Indeed, in the early years of the war, Hannibal consistently released prisoners of war who had fought as Rome’s allies, hoping the gestures would peel away Rome’s friends.²³

Following his victory at Cannae, Hannibal sent an emissary to Rome, expecting his spectacular victory would force the Romans to sue for peace.²⁴ Rome refused to meet him. Instead, Rome began preparing for a long war. Furthermore, though Hannibal hoped that Rome’s allies would flock to him, most did not: Rome lost the Samnites and some Apulians, but few others.²⁵ With the exception of Capua, no major city defected after Cannae, and many cities, most prominently Naples, rebuffed Hannibal’s overtures and weathered his assaults rather than betray Rome. Hanno, Hannibal’s domestic rival, summarized the situation like so:

‘But just what is this jubilation all about anyway […] since the battle at Cannae has meant the annihilation of Roman power, and since it is an established fact that all Italy is up in arms, has any people from the Latin league gone over to us? And, secondly, has any individual from the thirty-five Roman tribes deserted to Hannibal?’ When Mago [Hannibal’s ally in Carthage] answered ‘no’ to both questions, Hanno continued […] ‘So we have a war in which we are no further forward than the day Hannibal crossed into Italy.’ (Livy, 2006: XXIII.12–13)

To reinforce its psychological control of Italy, Rome turned to Capua. The strongest of the war’s defectors, Capua ranked second to Rome among the Italian cities.²⁶ Militarily, Rome did not need to take Capua: other cities, such as Tarentum, were more vulnerable and more militarily valuable.²⁷ But precisely this unnecessary risk was what Rome found so appealing: if Hannibal could not protect Capua, the second city in Italy, then he could not protect anyone. Like a sharpshooter aiming at the farthest target, Rome’s taking Capua would demonstrate Roman power conclusively.

Rome invested Capua as it did most any other city, weakening its defenses through starvation and the gradual collapse of morale. By the close of 212 almost half of Rome’s Italian legions lay encamped about the city, surrounding it with a double line of siege works and their own defensive perimeters.²⁸ Sensing the blow Capua’s loss would deliver his strategy, Hannibal marched on Rome. Rumors spread claiming that Rome had lost the siege, and citizens panicked in the streets at Hannibal’s approach. Yet the Senate refused to lift the siege of Capua, even at the risk of Rome itself, and finally Rome triumphed.²⁹ Rome took Capua, drove its aristocrats to suicide, and sold the population into slavery.

The fall of Capua terminated Hannibal’s hopes in Italy. One historian offers this assessment:

Rome persevered and would persist in replacing armies lost […] Symbolically and actually, all of this was epitomized by the wretched fate of Capua […] after Capua’s fall it became clear to all that Hannibal could not watch over widespread allies […] the war was far from over, but its outcome in Italy was all but decided. (O’Connell, 2010: 195–196)

To win, Carthage had to convince Roman allies to defect, but the fall of Capua revealed to all what would happen to those who did. Though he remained undefeated in open battle, no Italian city would afterward join Hannibal’s cause, and he would soon lose the war.

In retrospect, Cannae, a battle whose artistry military officers still study today, determined the outcome of the war far less than Capua, a straightforward siege. The Roman war effort – indeed, the entire Roman international system – depended on its ability to maintain alliances and raise armies from the Italian populace. Cannae did not significantly alter the perception of Rome’s ability to do either, while Capua communicated beyond all doubt Roman strength. Cannae was a great victory, but it was not symbolic. By contrast, Capua was a lesson to all Italian states: Hannibal can win on the battlefield, but he cannot protect his gains; defect and you will be destroyed. Rome undertook the riskiest battle of its Italian campaign as a signal, and that signal secured its victory.

Conclusion

This article opened with the question, why are some battles more significant than others? I have outlined one way a battle can gain an outsized importance: its symbolism. This symbolism, like war itself, is fundamentally political, and as such it speaks to a battle’s importance more than casualty ratios, tactical brilliance, or military value. A battle’s symbolism can come from many sources; one of the most important is its risk. Even small victories can become great symbols if unnecessarily dangerous, bold, or reckless. Capua, for all its banality, settled the war in Italy, while Cannae, for all its brilliance, signified – nothing.

In war, the primary source of information is the campaign, but our informational models of war often omit this source. Different types of states will pursue different paths to victory, and these different paths will signal their strength. In particular, by undertaking unnecessary risk, a state can communicate its strength. If a strong state takes an action which a weak state dare not, it clearly signals its capability. Of course, in actual war, the signals and risk a state can take are rarely so clear. But, while the signals will be mixed, they will exist. Stronger states will more often strive for more difficult, more dangerous objectives, and this tendency will drive at least some separation in assessments of power. I abstract from these complications to capture the essence of the signal, but I would note that in the real world far more signals are being sent than my stylized model would suggest.

Thus far the literature on bargaining during war tends to overlook a crucial signal – military strategy itself. Whether states fight and whether they win are not the only sources of information battles convey: where and how states fight can signify just as much. A symbolic victory can permanently change the strategies and morale of the belligerents, as in the Doolittle Raid. Sometimes, it can even end a war, not because it destroys a general’s enemies, but because it convinces his enemies (and his allies) that he is strong. Symbolism is not the only criterion separating important from unimportant battles, but it is one of them, and one that some of the most celebrated grand strategies have employed.

Footnotes

Replication data

The Online appendix can be found at .

Acknowledgments

Earlier versions of this article were presented at Princeton University and at ISA’s Annual Conference. I am grateful to David Carter, W David Clinton, Matteo Faini, Aaron Friedberg, Joanne Gowa, Michael McKoy, Adam Meirowitz, and especially Kris Ramsay, for their mentorship and feedback. I am also indebted to the editor of JPR and two anonymous reviewers for their extensive comments. Any remaining errors are my own.

Notes

ORCID iD

Richard Jordan

References

Admati

Anat R

Perry

Motty

(1987) Strategic delay in bargaining. Review of Economic Studies 54(3): 345–364.

Arreguín-Toft

Ivan

(2001) How the weak win wars: A theory of asymmetric conflict. International Security 26(1): 93–128.

Bennett

D Scott

Stam

Allan C

III (1996) The duration of interstate wars, 1816–1985. American Political Science Review 90(2): 239–257.

Caesar

Julius

(1951) The Conquest of Gaul. Trans. Handford

. New York: Penguin.

Craven

Wesley Frank

Cate

James Lea

, eds (1948) The Army Air Forces in World War II, volume 1. Chicago, IL: University of Chicago Press.

Crawford

Mark

(1999) Encyclopedia of the Mexican–American War. Santa Barbara, CA: ABC-CLIO.

Creasy

Edward S

Mitchell

Joseph B

(1964) Twenty Decisive Battles of the World. New York: Macmillan.

Davis

Paul

(1999) 100 Decisive Battles from Ancient Times to the Present. Oxford: Oxford University Press.

DeRouen

Karl R

Jr Sobek

David

(2004) The dynamics of civil war duration and outcome. Journal of Peace Research 41(3): 303–320.

10.

Fearon

James

(1995) Rationalist explanations for war. International Organization 49(3): 379–414.

11.

Filson

Darren

Werner

Suzanne

(2002) A bargaining model of war and peace: Anticipating the onset, duration, and outcome of war. American Journal of Political Science 46(4): 819–837.

12.

Frazier

Donald S

, ed. (1998) The United States and Mexico at War. New York: Simon & Schuster Macmillan.

13.

Fuller

JFC

(1954) The Decisive Battles of the Western World and their Influence Upon History. London: Eyre & Spottiswoode.

14.

Glines

Carroll V

(1988) The Doolittle Raid: America’s Daring First Strike Against Japan. London: Orion.

15.

Goemans

(2000) Fighting for survival: The fate of leaders and the duration of war. Journal of Conflict Resolution 44(5): 555–579.

16.

Goldsworthy

Adrian

(2000) The Punic Wars. London: Cassell.

17.

Kilcullen

David J

(2010) Counter-Insurgency. Oxford: Oxford University Press.

18.

Kydd

Andrew H

(2005) Trust and Mistrust in International Relations. Princeton, NJ: Princeton University Press.

19.

Langlois

Jean-Pierre P

Langlois

Catherine C

(2012) Does the principle of convergence really hold? War, uncertainty and the failure of bargaining. British Journal of Political Science 42(3): 511–536.

20.

Lawson

Ted W

(1943) Thirty Seconds over Tokyo. New York: Random House.

21.

Lazenby

(1978) Hannibal’s War: A Military History of the Second Punic Army. Norman, OK: University of Oklahoma Press.

22.

Leventoğlu

Bahar

Slantchev

Branislav L

(2007) The armed peace: A punctuated equilibrium theory of war. American Journal of Political Science 51(4): 755–771.

23.

Livy (2006) Books 21–30: Hannibal’s War. Trans. Yardley

. Oxford: Oxford University Press.

24.

Luttwak

Edward N

(1976) The Grand Strategy of the Roman Empire from the First Century AD to the Third. Baltimore, MD: Johns Hopkins University Press.

25.

Machiavelli

Niccolo

(2003) Art of War. Trans. Lynch

Christopher

. Chicago, IL: University of Chicago Press.

26.

Mackay

Christopher S

(2004) Ancient Rome: A Military and Political History. Cambridge: Cambridge University Press.

27.

Malkasian

Carter

(2006) Signaling resolve, democratization, and the first battle of Fallujah. Journal of Strategic Studies 29(3): 423–452.

28.

Maslowski

Peter

(1994) To the edge of greatness: The United States, 1783–1865. In: Murray

Williamson

Knox

MacGregor

Bernstein

Alvin

(eds) The Making of Strategy: Rulers, States, and War. Cambridge: Cambridge University Press, 205–241.

29.

Merrill

James M

(1964) Target Tokyo. Chicago, IL: Rand McNally.

30.

Montgomery

Evan Braden

(2019) Signals of strength: Capability demonstrations and perceptions of military power. Journal of Strategic Studies 43(2): 1–22.

31.

Nelson

Craig

(2002) The First Heroes: The Extraordinary Story of the Doolittle Raid – America’s First World War II Victory. New York: Viking.

32.

O’Brien

John Maxwell

(1992) Alexander the Great: The Invisible Enemy. London: Routledge.

33.

O’Connell

Robert L

(2010) The Ghosts of Cannae: Hannibal and the Darkest Hour of the Roman Republic. New York: Random House.

34.

Polybius (2010) The Histories. Trans. Waterfield

Robin

. Oxford: Oxford University Press.

35.

Powell

Robert

(2004) Bargaining and learning while fighting. American Journal of Political Science 48(2): 344–361.

36.

Powell

Robert

(2006) War as a commitment problem. International Organization 60(1): 169–203.

37.

Ramsay

Kristopher W

(2008) Settling it on the field: Battlefield events and war termination. Journal of Conflict Resolution 52(6): 850–879.

38.

Ramsay

Kristopher W

(2017) Information, uncertainty, and war. Annual Review of Political Science 20(1): 505–527.

39.

Reiter

Dan

(2003) Exploring the bargaining model of war. Perspectives on Politics 1(1): 27–43.

40.

Schelling

Thomas C

(1963) War without pain, and other models. World Politics 15(3): 465–487.

41.

Schultz

Duane

(1988) The Doolittle Raid. New York: St Martin’s.

42.

Slantchev

Branislav L

(2003) The principle of convergence in wartime negotiations. American Political Science Review 97(4): 621–632.

43.

Slantchev

Branislav L

(2004) How initiators end their wars: The duration of warfare and the terms of peace. American Journal of Political Science 48(4): 813–829.

44.

Smith

Alastair

(1998) Fighting battles, winning wars. Journal of Conflict Resolution 42(3): 301–320.

45.

Smith

Alastair

Stam

Allan C

(2004) Bargaining and the nature of war. Journal of Conflict Resolution 48(6): 783–813.

46.

Summers

Harry G

Jr (1982) On Strategy: A Critical Analysis of the Vietnam War. New York: Dell.

47.

US Department of Defense (1946) United States Strategic Bombing Survey Summary Report (Pacific War). Washington, DC: United States Government Printing Office.

48.

Wagner

R Harrison

(2007) War and the State: The Theory of International Politics. Ann Arbor, MI: University of Michigan Press.