CHAPTER II

Modeling Combat and Sizing Forces

IS IT POSSIBLE to make meaningful estimates of how wars will unfold? The most important result to try to predict is, of course, the winner. Even in situations where that might be rather foreseeable, as in the two U.S.-led invasions of Iraq in the last two decades, gauging the likely duration of the conflicts—and the likely casualties that will ensue to participating armies and proximate civilians—is of great interest when possible.

An important related question is: Can we make meaningful estimates of what force package would be adequate to prevail, and prevail decisively, in a proposed conflict? This question gets to the heart of force planning and defense budget analysis, for the United States and also for other countries. It should also influence decisions on whether to enter into war in the first place, for situations in which there is a real choice of when or whether to fight.

In the abstract, predicting outcomes in war is extremely hard. As Australian historian Geoffrey Blainey notes, for example, “when nations prepare to fight one another, they have contradictory expectations of the likely duration and outcome of the war. . . . it is doubtful whether any war since 1700 was begun with the belief, by both sides, that it would be a long war. . . . No wars are unintended or ‘accidental.’ What is often unintended is the length and bloodiness of the war. Defeat too is unintended.”¹

There is a reason so many people have been so unsuccessful in predicting the course of armed conflict. It is because war depends greatly on variables that are very hard if not impossible to quantify, such as the quality of leadership, the effectiveness of any surprise, and the performance of new weapons systems or military operational concepts not previously tested in battle (and hence not well understood in advance of battle).

According to historical data sets, even if one country or alliance is clearly strongly than another according to various military metrics such as overall manpower or combat equipment inventories or a combination of both, it is very hard to make high-confidence predictions about which side will win. Brookings senior fellow Joshua Epstein compiled evidence from the U.S. Army Concepts Analysis Agency published in the 1980s that made this point clearly. The Army’s Concepts Analysis Agency had gathered data on many past wars, and used that data to compute overall scores on the capabilities of attackers and defenders through a process described in more detail in the following. It turned out that for countries with roughly comparable military capabilities, with neither side having more than a 50 percent edge against the other in military inputs, the attacker won 142 out of 230 battles, or 58 percent of the total.² There was no tendency towards stalemate (only 7 percent of all cases). Nor was there a major inherent advantage for the defender (who prevailed in just 35 percent of all cases). Benefiting from tactical surprise and innovative concepts about how to fight, the attacker usually won, but no other broad conclusion could be reached from the dataset.³

Moreover, even though the attacking country usually won, it was far from assured of doing so. The attacking side won only 63 percent of the time when having an estimated advantage between 50 percent and 200 percent over its opponent, and 74 percent of the time when having an even greater edge.⁴ So even when military balances seem to clearly favor one side, outcomes are hard to forecast. This fact helps explain arms race dynamics. It is not easy to measure a military balance accurately. Correlations of forces always depend on specifics of when and where a war is fought. In addition, substantial initial advantages are needed to provide any real confidence about the likely outcome. In such a situation, it is no surprise that rival countries with grievances between them sometimes enter into escalating spirals of military buildups.

The difficulty of measuring balances accurately, and the need for a substantial advantage to provide any confidence of victory, help explain why the United States is so bent on having a substantial military edge in any battle. (As noted in the following, its advantages in recent conflicts have, when quantified, typically been three or four or five to one against likely foes without even counting airpower.) Few seriously doubt the likely outcome for such invasions by an overwhelming power against a far smaller, weaker, and less sophisticated power—at least when the outcome is defined in terms of who “captures the flag” or winds up in control of the capital city and major infrastructure. But that is only because the disparity in capabilities is so great. In such cases of extreme mismatch, it is even possible to hazard estimates about the likely duration and likely casualty levels from battle. Even if predictions are naturally imprecise, they can often predict the order of magnitude correctly (that is, actual results will be within a factor often of estimated results, and often within a factor of two or three). It has been recognized for years that expected casualties are generally an important consideration when Americans make decisions about whether and how to go to war, meaning that predictions about such matters can be important on policy grounds.⁵

Of course, as the case of Iraq also underscores, predicting the outcome of classic large-scale military engagements is one thing. Predicting casualties—or even the winner—in counterinsurgency missions and “irregular combat” is something else. So if the business of predicting the results of war is very hard—with errors of 100 to 200 percent amounting to relatively good results, and with far larger inaccuracies common when foes are nearly evenly matched or when battle involves guerrilla, terrorist, and urban operations—why bother with the business of modeling combat at all?

One reason is that, in science as well as other disciplines, understanding what we do not or cannot know is important in its own right. If a large body of human experience strongly suggests that warfare is an inherently risky and unpredictable business, the onus will be on policymakers who propose engaging in warfare to explain why it is necessary—and to explain how they have taken every reasonable precaution to minimize the chances of being surprised by the course of events. A good case in point is the Bush administration’s decision to engage in what some have called a “war of choice” to overthrow Saddam in 2003—and even more, its decision to do so with smaller forces than recommended by military officers or other analysts, and without a well-developed plan for keeping the peace and restoring order after Saddam was toppled. Greater care in preparation, and less confidence about the probable course of battle, would have been appropriate.

Of course, modeling does not always produce pessimistic forecasts. But it should at least explain and underline the types of assumptions needed to make a certain prediction. In preparing their devastating 1967 surprise attack on Arab air forces, Israeli planners realized that they might be able to achieve a remarkable success, based on a sense of how accurately bombs could be dropped, how easily enemy radar could be evaded, and how vulnerable airplanes and airfields would be to ordnance of a given size and explosive power. But they also had to appreciate how many things would have to go right, in terms of the timing and execution of the attack, to make it succeed. That helped focus them on making preparations very carefully. To take another example, when my former colleague at the Congressional Budget Office, Lane Pierrot, presciently argued in 1990 that it was possible U.S. aircraft loss rates in Operation Desert Storm would be as low as peacetime training rates, she was making the sophisticated yet simple argument that American countermeasures and flying practices might be good enough to counter much of the Iraqi air defense network. Pierrot foresaw that American airpower might render the latter air defenses largely powerless except when coalition aircraft had to fly low to go after certain targets, or penetrate particularly dense air defense bastions. But Pierrot also realized that losses could be much higher; she wound up accurately prognosticating the results in part because she did not seek to be too accurate, but focused instead on establishing plausible lower and upper bounds for expected losses. This is an important lesson about the proper way to model combat.

A second reason to model is that we actually cannot avoid it even if we wish to. As Joshua Epstein has convincingly argued, whether we “model” mathematically and systematically, or anecdotally and impressionistically, everyone who forms an opinion on whether a given war should be fought is in effect predicting its outcome or at least its plausible range of outcomes. In other words, everyone is operating from some image, some set of expectations, of the likely course of battle, whether precise or not. Otherwise, there would be no way to decide if a given war should be fought in the first place, or if 100,000 or 500,000 troops might be needed for it, or if the nation’s industry should be directed to prepare for a long struggle and if a general mobilization might be required.⁶ The issue is not really whether we try to find a simplified construct for predicting battle outcomes—all of us do; in fact, all of us must. The issue is whether we choose to employ impressionistic and purely subjective “modeling” or a more rigorous and formal approach. The advantage of formal modeling is that it requires one to make assumptions explicit, and justify them as well as possible using historical, technical, and operational data.

Elaborate computer simulations are not always the best way to model combat. To be sure, algorithms like the Institute for Defense Analyses’ TACWAR are always worth consulting. They explicitly simulate the use of multiple weapons over complex terrain, in variable weather conditions, and with numerous possible assumptions about the performance of air-power, logistics systems, and other key determinants of battlefield outcomes. Even these types of models generally have limits on their ability to mimic reality, however. TACWAR has in the past only allowed for a simple front line where armies encounter each other, rather than a dynamic and complex battlefield, and has limited the user’s ability to make various assumptions about the possible performance of different weapons.⁷ For these and other reasons, the complex models do not always produce the best predictions. Simpler calculations about the likely course of Operation Desert Storm in 1991 done by numerous outside analysts were generally more accurate than sophisticated Pentagon computer runs. While virtually all were too pessimistic, outside analysts generally estimated that U.S. fatalities would total around 1,000 whereas more elaborate models reportedly projected American deaths up to several times as numerous. (Actual combat losses were 146, and including all phases of the operation nearly 400 Americans lost their lives.)⁸

This chapter employs several different simple frameworks for modeling combat. All involve some amount of basic arithmetic. In no case should the use of math distract the reader from the need to critically evaluate assumptions, to think hard about where calculations may go wrong, and to remember how often warfare surprises us.

The discussion begins with fairly classic and simple models of traditional combat, starting with Lanchester’s equations. It then considers other models, some embodied in simple formulas and others best understood as a systematic way of thinking through a problem or scenario. In other words, the mathematics are not always paramount, and they are not always neatly condensed into one or two simple algorithms. After Lanchester, slightly more complicated models for armored combat are discussed. Then, other types of ground combat—infantry war, urban war, counterinsurgency—are considered together. The focus then changes to naval combat, including a scenario for amphibious assault and another for a blockade operation at sea. The old-fashioned (and hopefully obsolete) matter of nuclear exchange calculations is then discussed. The chapter concludes with a framework for analyzing progress in counterinsurgency operations like those in Iraq and Afghanistan.

LANCHESTER’S EQUATIONS

Although they are somewhat simplistic, the Lanchester equations are a good place to start in understanding how war simulations are conducted and combat models created. Formulated a century ago by a British engineer who gave the equations their name, they have the advantage of simplicity—since war was, on balance, somewhat simpler then, if still inherently hard to predict.

Lanchester equations come in several forms, but two are of particular note: those for direct fire (like rifles aimed at specific individuals) and indirect fire (such as artillery fired into a broad area where enemy forces are known to be located even if they are not individually visible and targetable). Of course, modern war has elements of both types of fire, so any sophisticated model would need to combine their effects (accounting as well for a variety of types of weapons, for complex terrain, for command and control, for maneuvers, and so on). But some of the basic dynamics are easier to see if combat is first simplified and abstracted into these two simple forms.

Direct-Fire Equation

Imagine two rows of eighteenth-century soldiers lined up firing at each other. Assume they are not quite shoulder to shoulder, but instead have some spacing between them (making it very unlikely that one soldier will be hurt by a weapon fired at another soldier). Assume further that there is no issue of concealment; all soldiers on both sides are within sight, and weapons range, of each other.

This model of combat may be a relatively good way to understand naval gunfire exchange at sea (more on that appears in a subsequent part of this chapter), certain types of aerial combat in the skies, and in general warfare where there is little subtlety or complexity to the battlefield—enemy forces see each other and try to destroy each other in a fairly straightforward exchange of gunfire.

With this image of combat in mind, and two armies represented by A(t) and B(t), the basic mathematics of the Lanchester equation are simple to derive. (They involve a small amount of very basic calculus; those unable to follow it need not worry since the result can be understood just through arithmetic.) A(t) and B(t) are the total numbers of soldiers on each side still unwounded and capable of fighting at a given time. Clearly, A(t) will diminish faster the larger the number of enemy soldiers on the other side and the more effective that side’s weapons. Weapons effectiveness, which can be simplified as the multiplicative product of the rate of fire, the accuracy of the weapon, and the lethality of the weapon, is condensed into a single term for each side (represented by a and b, respectively, for army A and army B). In mathematical form, assuming a simple linear relationship, this can be written as follows (those uninterested in the derivation or put off by the calculus can simply skip a few lines to the actual formula):

dA(t)/dt = –bB(t)

Similarly, for B(t), we have:

dB(t)/dt = –aA(t)

In other words, A’s forces decline faster to the extent that B has a larger army firing more lethal weaponry at them, and vice versa.

By the chain rule, dA(t)/dt can also be written as:

dA(t)/dt = [dA(t)/dB(t)][dB(t)/dt]

Substituting the term on the right into the first equation, we get:

[dA(t)/dB(t)][dB(t)/dt] = –bB(t)

Further substituting for dB(t)/dt from the second equation in this list yields:

[dA(t)/dB(t)][–aA(t)] = –bB(t)

Rearranging terms and canceling out the respective minus signs we get:

a[A(t)][dA(t)] = b[B(t)][dB(t)]

Integrating, and expressing for the time t = T, we get:

A2(T) –A²(initial) = [b/a][B²(T) –B²(initial)]

In other words, the square or second power of A(t) is on the left hand side, for two different times, the initial time at the beginning of the battle, and a time T later in the course of events. The right side of the equation has a similar term for B, times the ratio of the typical weapons effectiveness for side B to that for side A.

As one example, assume army A starts with ten soldiers and army B with twenty (at time t = 0), and assume weapons effectiveness to be the same on both sides. How many soldiers will army B still have standing when A’s force has been wiped out (at time t = T)? This boils down to solving the following arithmetic problem:

0² –10² = B²(T) –20²

10² is of course 100 and 20² is 400, so this becomes:

 

–100 = B²(T)–400

  300 = B²(T)

 B(T) = √300 = 17 (roughly)

 

In other words, B loses just three soldiers in annihilating all ten of A’s soldiers. The Lanchester direct fire equation gives a great premium to numerical superiority, all other things being equal. (Every soldier in B’s army on average faces only half the fire of one soldier on the other side, while every soldier in A’s army by contrast is being shot at on average by two soldiers from B, explaining why A is at such a disadvantage in this engagement.)

Indirect Fire

Of course, at least two things are unrealistic about the scenario behind the direct-fire Lanchester equation. First, at least for all eras since the eighteenth century, enemy forces will take cover to get out of the way of weapons. Second, many weapons are fired in a way to barrage an area rather than directly strike a specific individual or vehicle. Pure musket-fire exchanges no longer tend to occur in warfare.

The indirect-fire version of the Lanchester equations corrects for these problems. It does so at the price of going to the other extreme and eliminating any role for direct-fire weaponry. But it is still instructive to see how the math changes, and how the predicted battlefield results can also therefore change. If nothing else, this helps us appreciate the stark differences between battlefield dynamics for wars dominated by direct fire and those dominated by indirect fire. (Capturing both direct and indirect fire in a single Lanchester equation is more realistic, but also much more complicated mathematically. In fact, when trying to capture such complex dynamics, I do not employ Lanchester equations here, but other types, as discussed further on in this chapter.)

In Lanchester’s indirect fire equations, the likelihood that one’s forces will suffer losses becomes proportionate to three terms. Two are as before, the effectiveness of the enemy’s weapons and the number of enemy forces. The third term is the density of one’s own forces on the battlefield. Having more people means having more targets, increasing the chances that an enemy weapon fired more or less randomly at a broad area will, by chance, strike soldiers in that area upon detonation. So we have:

dA(t)/dt = –bB(t)A(t)

and

dB(t)/dt = –aA(t)B(t)

Again, using the chain rule employed earlier:

dA(t)/dt = [dA(t)/dB(t)][dB(t)/dt] = –bB(t)A(t)

And then substituting the second equation for dB(t)/dt gives:

[dA(t)/dB(t)][–aA(t)B(t)] = –bB(t)A(t)

Rearranging and simplifying gives:

a[dA(t)] = b[dB(t)]

A(T) –A(initial) = [b/a][B(T) –B(initial)]

With this formulation, there is much less benefit to numerical superiority, as having more troops on the battlefield gives one more shooters but also provides the enemy with more targets. As such, the effectiveness of weaponry becomes just as important mathematically as a quantitative advantage in soldiers, something that was not true in the direct-fire equation. Mathematically, the number of one’s own forces is taken only to the first power, and not squared, as a result.

As one simple example, if A and B have equally effective weapons, and A starts with ten soldiers while B begins with twenty, they will lose personnel at the same rate. Army A will have five people left when B has fifteen left; A will run out of soldiers when B has ten still standing. Again, B wins, but less dramatically. And in a situation where A has somewhat better weapons than B, say 2.1 times as good, it could win. By contrast, in the direct-fire equations, it would need to possess a huge (fourfold) advantage in weaponry in order to compensate for its fewer forces and prevail in the engagement.

PREDICTING AIR-GROUND COMBAT

Simple models for predicting the course of heavy armored combat can best be explained with reference to a specific example such as Iraq. That is because a couple of simple formulas tend to be used together in these, and it is easiest to see how the overall calculation is done with reference to a concrete example rather than abstractions.

Prior to Operation Desert Storm in 1991, a number of scholars, using models and databases developed largely for assessing the NATO–Warsaw Pact military balance during the Cold War, estimated the losses likely to result in a war to expel Iraqi forces from Kuwait. Virtually all these estimates were too high, but they were also generally more accurate than those produced by the Pentagon before the U.S.-led war against Iraq began. Indeed, they were virtually all correct in predicting a short decisive conflict in which U.S. casualties would be far less than American losses in the Vietnam or Korean wars. In that sense, the flawed estimates were still useful.⁹

The Kugler–Posen and Epstein Models

At least two main families of models were developed during this time period in the open literature. As I would define and group them, they might be termed the Kugler–Posen “attrition-FEBA expansion” model and the Epstein adaptive dynamic model.¹⁰ Both are more sophisticated than the famous century-old Lanchester equations, which as noted require simplifying assumptions about the nature of weaponry that apply much better to eighteenth-century musket fire, nineteenth-century battleship duels, or World War I artillery exchanges than to the modern battlefield.¹¹ They are much less sophisticated than the detailed, and classified, computer models such as “TACWAR” and “Janus” used by the Pentagon community to predict combat outcomes. But they do offer simplicity and accessibility.¹² And again as noted earlier, they also have every bit as good a track record in recent times of predicting combat outcomes. What they may lack in detail and exactitude they tend to make up for by requiring a user to think pragmatically, historically, and intuitively about the modeling enterprise—rather than running the risk of getting lost in the math, or being overly impressed by the internal machinations of the complex computer programs.

The first of these unclassified and relatively simple models, developed by Richard Kugler (of National Defense University in Washington) and Barry Posen (of MIT), was optimized for a war of attrition in which NATO was presumed to be on the defensive. It is based on the assumption that a military of sufficient size can hold a front of a given length against all-out enemy assault. Provided that the defender can reinforce its losses and maintain an adequate “force-to-space ratio” as the forward edge of the battle area (or FEBA) evolves, it should be able to hold the line and protect its territory.¹³ Posen later applied the model to the U.S. plan to liberate Kuwait from Iraq, based on the assumption that such a war would resemble a NATO–Warsaw Pact confrontation in the types of weaponry and tactics employed. In that case, he was NOT assuming that the Iraqis could hold their positions robustly, but used the equations primarily to calculate relative casualty levels. There was no inconsistency in this approach, since Iraqi forces were unable to reinforce well during an attack, meaning they could be worn down quickly in an attempted breakthrough sector.

The second model, developed by Joshua Epstein at the Brookings Institution, has numerous similarities with the Kugler–Posen framework, but it challenges the idea that a sufficient “force-to-space” ratio ensures a viable defense. That ratio is typically cited as one armored division equivalent (ADE) per twenty-five kilometers of front according to proponents of that approach, though some estimates suggest an ADE can cover twice as much frontage—underscoring the imprecision of this rule of thumb.¹⁴ Epstein also rejects the popular idea that a defense can fend off an attack provided that the offense does not achieve a 3:1 force advantage in the sector of attempted breakthrough.¹⁵ In fact, as he convincingly shows, and as databases like those of the late Trevor Dupuy and the Army’s Concepts Analysis Agency demonstrate as well, attackers have often succeeded historically when simply equal to (or even smaller than) the defense.¹⁶

Epstein’s model also allows for the possibility that a defender might withdraw in order to buy time, improve its position, or slow the pace of battle—thereby slowing the rate at which casualties are incurred. In other words, geography and the movement of forces are still part of the analysis, but in a different way, tied to casualty rates rather than force-to-space ratios.¹⁷ The differences in these methods can be quite significant in a given case.

For a war in which breakthrough operations are successful, and in which organized withdrawal is not a major factor, the Posen–Kugler and Epstein approaches can be collapsed into a single approach. This combined approach at least gives a sense of relative casualties on the two sides. That is the method followed here.¹⁸

In fact, the methods have many commonalities. Both focus on armored divisional equivalents and their heavy weaponry as the main variables in their associated equations. “Armored division equivalents” not only reflect the quantity of armored formations, but also their equipments’ quality. They are defined such that a modern U.S. heavy division equals 1.0 ADEs.

The models also specifically incorporate a role for ground-attack aircraft in the close-air support role. They are assumed to be capable of dropping a given number of munitions per sortie, and flying a given number of sorties per day (with a given probability of being shot down on each mission). Each munition is assigned a “kill probability” to reflect the likelihood it will strike and destroy a major ground vehicle. Knowing the number of such ground vehicles per division, one can translate the attrition from aerial attacks into armored division equivalents, thereby linking the ground and air wars conceptually and mathematically.

In modern war, of course, airpower is used against many targets besides vehicles in the field. That was not only true in the massive city bombing campaigns of World War II, and the “carpet bombing” of Vietnam, but also in Operation Desert Storm, the Kosovo war, and other cases.¹⁹

In Desert Storm, for example, while just over half of all strike missions were against fielded Iraqi forces, the remaining 43 percent focused on airfields, SCUD missile launchers, surface-to-air missile sites and other air defense infrastructure, strategic lines of communication including bridges and port facilities and train lines, military industry, suspected weapons of mass destruction sites, telecommunications, electricity grids, oil assets, and leadership targets. A total of about 1,000 U.S. attack aircraft (plus another 250 or so from allies such as Saudi Arabia and Britain) dropped around 200,000 munitions, just under 10 percent of them precision-guided.²⁰ In NATO’s 1999 air war against Serbia, Operation Allied Force, roughly 7,600 strike missions were flown against fixed targets and 3,400 against mobile targets. A total of about 1,000 NATO aircraft were ultimately employed in the conflict (including strike planes, air superiority fighters, and support aircraft), dropping some 28,000 munitions, of which about 7,000 were precision-guided. Only two fixed-wing aircraft were lost over Serbia, another malfunctioned and crashed in the Adriatic, and two Apache helicopters crashed over Albania, causing America’s only two fatalities of the war.²¹

In the models in question, however, such strategic bombing and “battlefield air interdiction” is viewed as preparatory or exogenous; airpower enters into the mathematics only through its effects on ground power (and possibly through its effects on either side’s ability to reinforce key positions).

Each of the two models assumes that the attacker drives the pace of battle and that it is able to adjust the pace of combat up to a certain reasonable maximum. That maximum rate of attacker losses is usually 1 percent to 5 percent per day for a division-sized force, with the lower numbers more typical of protracted fighting and the higher end of the range occasionally attained in intensive combat. This is a historically based figure, consistent with most of the experiences of World War II battles, subsequent Arab–Israeli wars, and other conflicts.²² A user of either model begins by specifying an assumed daily loss rate for the attacker in the ground battle within this range. Epstein’s model allows the defender a say in the intensity of battle, too, by allowing the defender to stage a withdrawal that, as noted, reduces the assumed casualty rates on both sides.

Both models then require the user to estimate an “exchange rate.” This is a proportionality factor linking the losses of the attacker to those of the defender. This exchange rate reflects first and foremost the quality of troops and equipment on each side of the war. Losses are expressed in terms of armored divisional equivalents (or ADEs), the coin of the realm in these models. Human casualties can then be inferred once ADE losses have been estimated. (Depending on which country’s military is at issue, a division usually has 10,000 to 18,000 soldiers, and usually one to three actual divisions are in any given armored divisional equivalent. This is because less sophisticated militaries often need to put two or three of their divisions on the battlefield to create the effectiveness of just one modern Western division, the standard by which an ADE is defined.)

Applying Models to Operation Desert Storm

How did these various models do in estimating the outcome of Operation Desert Storm? It is worth working through the numbers for an important example like this for three reasons. First, doing so shows how the different pieces of the model (notably the ground war component and the air war component) fit together methodologically. Second, doing so provides useful practice with a concrete case. Third, doing so helps one see the likely limits on accuracy—even for a relatively successful case of modeling—but also its potential, assuming that expectations about what modeling can deliver are kept in check.

Both models were used to predict rapid decisive victory by coalition forces, with considerably higher casualties on the Iraqi side than the American side, and in that regard Posen and Epstein were both correct. More specifically, Posen forecast weeks of combat and 4,000 to 11,000 coalition casualties to liberate Kuwait (including dead and wounded).²³ Epstein predicted weeks of combat as well, and a slightly broader casualty range of 3,000 to 16,000 (again, dead and wounded combined).²⁴ Both made calculations based on the premise of attrition warfare (albeit shortlived attrition warfare), after relatively short air campaigns, given what was known about likely Pentagon war plans at the time. The assumption of attrition warfare was not altogether incorrect, at least in the opening hours of combat, for U.S. Marines and associated forces who penetrated Iraqi defenses and drove towards Kuwait City. It was incorrect for the forces led by the Army’s Seventh Corps, which executed the famous “left hook” to the west of Iraqi defenses, outflanking Iraqi forces in their initial movements, though engaging in occasional combat with Saddam’s military thereafter.²⁵

Press reports suggested that the Pentagon was prepared for 30,000 or more casualties in Operation Desert Storm, even though it presumably did have access to detailed battle plans when making its predictions.²⁶ For such estimates, one would presume that 15 to 20 percent of all U.S. casualties would have resulted in deaths and the rest in wounded personnel, given historical trends (that account for the improving benefits of modern medical care).²⁷

In the actual event, losses were less than forecast. By official count, a total of 382 Americans died in the southwest Asian theater in the course of Operation Desert Shield, which began in August 1990, and Desert Storm, as that operation was renamed in January 1991. That count includes prewar and postwar accidents and other non-hostile acts. A total of 147 U.S. troops died in combat. Thirty-five were killed accidentally by so-called friendly fire. Others died in accidents of various types, on and off the immediate battlefield. About 500 additional Americans were wounded.²⁸ Considering allied forces as well, and using round figures, the coalition suffered about 240 combat deaths, some 500 deaths over the course of the entire operation from all causes, and about 1,500 casualties including killed and wounded.²⁹

How good were the prewar estimates, and what do the inaccuracies tell us about the value of trying to predict casualties? On the whole, these casualty estimation efforts were rather successful despite their inaccuracies, especially for the broad policy points they implied—that war would be decisive, victorious, and not very bloody by the standards of past major conflicts (yet hardly casualty-free).

With a clearer indication of how long the air war would last prior to the ground campaign, the preceding models could have done an even better job of estimating likely casualties in Desert Storm. And after the fact, it is also possible to adjust various parameters to account for the better-than-expected American performance and worse-than-expected Iraqi performance—providing more accurate tools for modeling any subsequent similar conflict (such as the invasion phase of the 2003 war).

On the latter point, Stephen Biddle has enumerated many of the basic mistakes the Iraqis made. Among other things, they failed to post advance guards near trench lines and failed to remove dirt from the vicinity of those trench lines to keep the locations of dug-in forces secret.³⁰ While the United States had reconnaissance technologies, such as the Joint Surveillance and Target Attack Radar System (JSTARS), that made it much easier to detect moving Iraqi vehicles, and while infrared detectors helped it find vehicles at certain hours of the evening in particular, it was often unable to know where Iraqi units were through theater-wide surveillance.³¹ As such, Iraqis often could have gotten in the first shot, if they had properly exploited their defensive advantages.

In addition, American forces benefited from their supporting superstructure—intelligence, communications, equipment maintenance, and logistics support—even more than expected. The models, focused as they are on individual armored combat units and their traditional weaponry, and dependent on past combat data to generate battlefield performance parameters, do not tend to highlight such capabilities. These facets of modern war give an even greater benefit to a military like the U.S. armed forces, capable as it is of competently assimilating them all into the way it fights, and confer an even greater disadvantage on a country like Iraq that fails to understand or counter them.³²

High technology, particularly the ability of U.S. airpower to prepare the battlefield for more than a month before the ground war, also played an unanticipated role. For example, the tactic of “tank plinking,” in which laser-guided bombs were dropped on Iraqi armor (often in the early evening, when the desert sands cooled more quickly than Iraqi armor, revealing the locations of the latter to infrared sensors), was only developed in the course of the war. It could not have been easily foreseen in a combat prediction done before the war began. The ability of coalition aircraft to undertake that and other effective tactics from high altitude, out of range of Iraq’s man-portable surface-to-air missiles, was also not foreseen—even by war planners, who had coalition pilots fly low for the first days of battle. More generally, American military equipment turned out to be even better than expected, compared to Soviet weaponry like that fielded by the Iraqis.³³ As a result of all these factors, the American-led victory over Iraq was far more decisive than Israeli victories in previous wars against Syria, Jordan, and Egypt.³⁴

These factors can be adjusted to make future predictions more accurate. The ability of coalition forces to wage an air war indefinitely prior to any ground assault can be reflected in how the models are used, as can the high lethality of modern air-to-ground ordnance.³⁵ Superior American fighting capability and poor Iraqi competence can be reflected in a lopsided “combat exchange ratio” that further amplifies the adjustments already made due to varying equipment quality. The U.S. ability to stay out of range of much Iraqi fire, at least on the open battlefield, can be reflected in a much lower daily attrition rate for the attacker than usually assumed.³⁶

One can get the gist of this without wading through complex calculations. Considerable uncertainty still surrounds the issue of Iraqi losses in Operation Desert Storm. But Iraq appears to have lost roughly 1,100 to 1,400 tanks, about 800 armored personnel carriers, and 1,000 to 1,500 artillery tubes during the air war. That makes about 3,300 major pieces of weaponry. (In a standard U.S. division in the modern era, there have been about 1,200 such major weapon vehicles per division.)³⁷ It lost another 1,000 to 1,200 tanks, about 700 armored personnel carriers, and 1,000 or more artillery tubes during the ground campaign. These losses came out of initial Iraqi assets of up to 4,000 tanks, 3,000 artillery tubes, and 3,000 armored personnel carriers in the Kuwaiti theater (as well as about 340,000 personnel).³⁸ Iraqi personnel casualties are even more uncertain, but probably numbered in the low tens of thousands. (Somewhat more than 2,000 Iraqi civilians are also believed to have died in the course of the conflict.)³⁹

Knowing this information, we can redo the math to “predict” what happened in Operation Desert Storm more accurately. For the air war, equipment losses resulted from a total use of about 10,000 precision-guided air-to-ground munitions (PGMs) including Maverick and Walleye air-to-surface missiles and laser-guided bombs, as well as from ground fire.⁴⁰ Mathematically, the air war is very simple to represent. Assuming a kill probability of about 65 percent per weapon launched (modern norms to that point were typically perhaps 5 to 10 percent),⁴¹ we get:

[10,000 (weapons)][0.65 (kill probability)] = 6,500 destroyed vehicles

Of that total of 6,500 vehicles, assume for simplicity that half were main combat vehicles and the other half support vehicles like trucks (a reasonable assumption for most militaries).⁴² This translates into about 3,300 Iraqi combat vehicles “predicted” to be lost.

Following the Kugler–Posen and Epstein frameworks, one can then proceed as follows for understanding the ground war. Coalition forces had the equivalent of roughly ten armored divisions in place prior to the outbreak of hostilities (that is, ten ADEs).

By contrast, Iraq’s forces included about 2.8 equivalent divisions after the effects of the air war, and once the effects of poor Iraqi technology are factored in.⁴³ That number is calculated by taking the total number of major Iraqi weapons vehicles, which equaled about 10,000 before the air war, subtracting out the 3,300 vehicles destroyed during the air war (leaving some 6,700 in place), dividing that figure in turn by 1,200 armored vehicles per division (resulting in an estimate of 5.6 total divisions), and then adjusting the 5.6 divisions to account for the lower quality of Iraq’s divisions. That last step means dividing by two in this case, resulting in an Iraqi armored “score” of roughly 2.8 ADEs.⁴⁴

For the ground war, one now performs an iterative calculation day by day. Assume that coalition ground forces would lose 0.1 percent of their strength per day, a relatively low total by historical norms for intense combat.

Assume further that they would benefit from roughly a 30:1 combat exchange ratio in terms of “armored division equivalents.” It is this step in particular that benefits from hindsight; prior to Desert Storm, the best proxy for a U.S.–Iraqi conflict seemed to be the various Arab–Israeli wars in which Israel often destroyed three to five times as much Arab equipment as it lost from its own military (whether on offense or defense).⁴⁵ The fact that the United States could do much better than Israel had done was quite surprising, and helps explain why most modelers overestimated American losses.⁴⁶

On day one, coalition forces would thus lose 0.01 ADEs (0.1 percent often ADEs). Iraq would lose thirty times as much, due to the 30:1 combat exchange ratio working against it, or 0.3 ADEs from ground fighting. That is:

 

Iraqi Losses = [U.S. Losses][Exchange Ratio]
= [0.01 ADE][30] = 0.3 ADEs

 

On day two, the losses would round to the same amount as for day one (they would actually be slightly less, but only 0.1 percent less).

On day three, again losses would round to the same totals, and the same would be true for day four (this is an unusual result due to the very low coalition loss rate).

In addition, there would be losses from the ongoing air war. If the pace of aerial attack remained roughly as before, the United States might be expected to use another 1,000 munitions during this phase of conflict, destroying 650 Iraqi vehicles in the process. That would translate into 0.27 division equivalents or 0.14 ADEs over the three days in question. (U.S. aircraft losses, while not insignificant, were very low and strategically inconsequential, given the ability of the American armed forces to replenish losses and the limited duration of the war. A total of thirty-eight coalition aircraft were lost and forty-eight damaged by enemy action, almost all from ground-based air defenses—with heat-seeking or infrared surface-to-air missiles as well as anti-aircraft artillery causing more than two-thirds of all attrition.)⁴⁷

So after four days, the approximate duration of the ground campaign in Operation Desert Storm, cumulative losses from the ground war would be 0.04 ADEs for the coalition and 1.34 ADEs for Iraq—meaning about 2.7 Iraqi divisions overall, or 3,200 pieces of heavy weaponry (and perhaps 6,500 vehicles including trucks and the like).

Assuming 18,000 soldiers per ADE, the U.S. losses translate roughly into some 700 casualties, of which about 100 to 150 might be expected to be fatalities and the rest wounded, based on past norms.

In the general case, each day of war could be more complex to model than indicated here. Reinforcements could arrive from outside of the theater, for example. Forces could move on and off the battlefield. Both sides, not just the United States, could use airpower. But for this particular conflict, the preceding math is fairly accurate.

It was reasonable to think that the results of Desert Storm, incorporated into the previous models, would allow relatively accurate predictions of the outcome in 2003. Indeed, they did—for the invasion phase of the war, that is. Most weapons used against armor—laser-guided bombs, Mavericks, and so on—were more or less unchanged relative to 1991.⁴⁸ By 2003, the United States had the all-weather satellite-guided joint direct attack munition (JDAM). But the GPS-guided JDAM weapon typically misses its target by five to ten meters, so it is not always sufficiently precise to strike armor.⁴⁹ Moreover, weather was not a severe handicap in Desert Storm, so adding all-weather capability to the U.S. PGM inventory might have been expected to make only a marginal difference under similar circumstances in the future.⁵⁰ As for the Iraqis, even if they corrected some of the mistakes they made in 1991, they could have been expected to make other mistakes.⁵¹ Moreover, U.S. forces could change tactics in the event that Iraq found a way to hold its own in a given type of firefight, fighting more at night or relying more heavily on attack helicopters or working harder to avoid Iraqi defensive positions rather than driving straight through them. Of course, this logic broke down when another type of warfare, insurgency combined with terrorism, was adopted by the Iraqi resistance.

MODELING URBAN AND INFANTRY WARFARE

Urban and infantry warfare are often even more challenging than heavy air-ground combat for several reasons. Enemy forces are interspersed among civilian populations that need to be spared as much as possible, on moral as well as strategic grounds, and complex terrain complicates matters as well.

For such combat, a modified and simplified version of the model of the late U.S. Army Colonel Trevor Dupuy can be useful. The Dupuy method does not include a specific means for incorporating the effects of airpower and geography, so in that sense it is less sophisticated than the Kugler–Posen and Epstein models. Its advantage is that it focuses on soldiers, not armored divisional equivalents, making it more useful for infantry combat in which armored formations are generally less central. It is to first approximation an infantry model rather than a heavy air-ground warfare model. As with the Kugler–Posen and Epstein models, it allows the user to modify input data to reflect the quality of each side’s troops and equipment. It is also informed by a very detailed dataset on past conflicts. In addition, it incorporates coefficients for a wide range of factors such as weather, surprise, and terrain that require subjective interpretation to employ, but that allow for more explicit consideration of these elements of combat than the other two models.⁵²

Dupuy’s methodology is a bit hard to follow in its detail, but sensible and logical in its main framework. I simplify it here somewhat. He begins by translating the number of troops fielded by each side into a total power figure, P. P is the product of the size of the fielded force with its quality. It is further modified to account for the degree of surprise achieved in the early days of battle (for an attacker) and to account for the benefits of any concealment, complex terrain, and prepared positions (for a defender).

Using these power figures to calculate relative casualties requires the use of detailed and lengthy tables that reflect Dupuy’s experiences with a wide body of combat data from many past battles. In essence, each side’s daily casualties are estimated to be the product of three main types of terms: total troop strength, multiplied by a daily maximum casualty rate, multiplied by a factor accounting for the power differentials between the two sides.⁵³

In mathematical form, the terms for the attacker and defender are:

  P_ATTACKER = (N_ATTACKER)(Q_ATTACKER)(S_ATTACKER)

P_DEFENDER = (N_DEFENDER)(Q_DEFENDER)(S_DEFENDER)

For the attacker, the first term is the number of its troops, the second term is the relative quality of its troops, and the third term represents situational factors of direct relevance to the nature of the attack (notably the ability to achieve surprise). For the defender, the quality term can be set equal to 1 for simplicity (meaning that the attacker term for quality will reflect the disparity of the two sides’ capabilities). As noted, the situational terms for the defender include the effects of terrain, prepared positions, and weather. Typically, the situational factor would range from 1 to 2, depending on the degree to which the attacker is aided by surprise or the defender by weather, terrain, and prepared defensive positions. (Each of the individual terms—for weather, for terrain, for the nature of prepared positions, can typically vary by up to 50 or 60 percent, meaning that its representation in a power formula might vary from 1 to 1.6 on average.) Dupuy’s books provide detailed estimates of what the various factors might be for different types of defensive positions, varying degrees of poor weather, and the extent to which the attacker achieves surprise.⁵⁴

The terms for daily casualties might then be (taking 1 percent losses as a “norm” and calculating variance around that norm for each side):

  C_ATTACKER =(0.01)(N_ATTACKER)(P_DEFENDER/P_ATTACKER)

  C_DEFENDER =(0.01)(N_DEFENDER)(P_ATTACKER/P_DEFENDER)

The simplicity of this equation (like all others considered here) is both its strength and its weakness. There is some arbitrariness in the calculation of each side’s power, but consulting Dupuy’s books gives some historical perspective as to how the qualitative factors can be reasonably estimated. Certainly Dupuy’s own personal track record was rather good in using historical analogy and instinct to estimate such factors. It can be harder for someone else with less experience trying to use his approach, but the extensive databases in his books make the process somewhat tractable.

The Dupuy method, with its focus on foot soldiers, seems best suited to infantry battle. Its lack of focus on both airpower and maneuver warfare would suggest it is less useful for modeling heavy combat. It is only fair to note, however, that Dupuy also applied his model to predicting Desert Storm casualties, with accuracy comparable to that of the other two models discussed before.⁵⁵ Also, for mixed cases involving heavy combat as well as infantry battle, it can be a useful tool. So it is applied in the following to two cases: the U.S. invasion of Panama in 1989, and the U.S. invasion of Iraq and subsequent effort to stabilize that country beginning in 2003.

Panama

In December 1989, U.S. forces overthrew Panamanian strongman Manuel Noriega and defeated his armed forces. About 22,500 American personnel participated. They included Navy Seals and Army Rangers and other Special Forces. They also included large numbers of the 10,000 American troops stationed in Panama, such as the 193^rd Infantry Brigade. Soldiers from the 82^nd Airborne Division, 7^th Light Infantry Division, and 5^th Mechanized Infantry Division also participated. The operation involved simultaneous nighttime airborne operations against twenty-seven objectives throughout the country. Special forces infiltrated key sites shortly before the airborne assaults to take down Panamanian communications and oppose any attempts by Panama to reinforce its forces under attack. The massive simultaneous assault against Panama’s 4,400-strong defense forces and its paramilitary forces of several thousand more personnel overwhelmed the latter, surprising them with its ferocity and coordination in the opening hours of battle. Twenty-three Americans died, as did about 125 Panamanian military personnel.⁵⁶ Some 200 to 600 Panamanian civilians died as well.⁵⁷

As explained previously, Dupuy’s approach begins with a calculation of the power of the two sides. With U.S. forces 22,500 strong, if they are accorded a quality advantage of 3:1 over Panama’s military, and assumed to enjoy a 20 percent benefit from surprise, multiplying these factors together gives them a power score of about 80,000. (In his own book, Dupuy uses a somewhat smaller force estimate and a somewhat larger quality advantage, and estimates U.S. power at 75,000.) For Panama, counting about 4,000 paramilitary forces as well as soldiers, it had about 8,500 troops available; the fact that they fought on the defensive and in complex terrain is assumed to give them a doubling of capability score, for a total power of almost 20,000.

So the U.S. power advantage would be about 4, as calculated here (which shows up as the inverse of 4, or 0.25, in the first equation that follows).

American casualties can then be estimated at (22,500 × 0.01 × 0.25) = 56 a day. Over two days, U.S. casualties would be about 112, with about twenty killed. For Panama, flipping over the power ratio term, we get (8,500 × 0.01 × 4) = 340. Over two days, Panama’s casualties are estimated at 680, with about 170 killed. These results track reasonably well with the actual outcome.⁵⁸

Operation Iraqi Freedom

Before the Iraq invasion of 2003, I used Dupuy-like methods to get a rough sense of how many casualties might be suffered in the course of the conflict, predicting that up to several thousand Americans could die in the struggle. My estimates have, as of this writing in 2008, sadly turned out to be more accurate than seemed initially likely. My roughly correct answer was obtained, however, for partially wrong reasons. That is, my calculations assumed a hard but brief urban fight rather than a protracted guerrilla campaign. These mistakes largely balanced each other out. To put it differently, my instincts about how the war might go were largely wrong, but by being forced to follow the rigor of a model, I wound up with reasonably accurate results nonetheless. That is a testament to the value of models; by forcing one to posit better and worst cases, they can lead to a healthier acknowledgment of uncertainty than many would allow relying solely on personal judgment and intuition.

Here’s how the calculation might go. The United States employed almost 200,000 troops in the initial invasion of Iraq—closer to a quarter million, perhaps, counting those in support roles in the broader combat theater as well. Assuming a quality advantage of 3:1 over Iraq’s military, and some limited benefit from surprise, the assembled U.S. forces would then have a Dupuy “power value” of about 800,000.⁵⁹ If only 100,000 Iraqi personnel are assumed to have fought the United States, with the rest essentially disbanding themselves as the war began, Iraqis would then have a power value of about 200,000 (assuming situational factors from fighting on the defensive that roughly doubled their combat power).⁶⁰ That makes the relative power term in Dupuy’s casualty equation about 0.25 for the United States and about 4 for Iraq.⁶¹

If we assume intensive combat, percentage losses per day might be as great as in a smaller conflict like Panama. (Normally, as Dupuy’s databases show, larger armies lose a smaller percentage of their forces per day, because large parts of them are well behind front lines at any given time. The relationship varies roughly as the square root of the size of the force—so if an army grows tenfold in size, it will expect to have roughly a threefold reduction in its proportionate loss rate.)⁶² So, according to this logic, the daily loss coefficient would remain 0.01.

U.S. losses would then be roughly (250,000 × 0.01 × .25) = 625 casualties per day, and Iraq casualties would be roughly (100,000 × 0.01 × 4) = 4,000 per day. For a short war lasting just three days, U.S. casualties could then be about 1,900, of which about 400 might be fatalities. That is how I calculated the lower bound estimate before the U.S.-led invasion of Iraq in 2003.

To estimate the upper bound, I assumed more Iraqis might fight—a quarter million, to be precise. I further assumed the invasion phase of the war could last ten days rather than just three. That led to a U.S. power advantage of just 1.6:1, and an estimate of 1,560 American casualties a day or 15,600 for the whole conflict. That in turn implied about 3,000 dead, the upper range of my estimate.⁶³

As of this writing in late 2008, casualties in Iraq have exceeded my upper bound. More than 4,000 Americans have now died in Iraq. In effect, I predicted far too high of a U.S. loss rate—but over far too short a period.

To put it differently, even though I did not do so, the Dupuy equations can be used to model a protracted insurgency/counterinsurgency campaign by adjusting the daily casualty rate appropriately and then playing out the calculation over a long period of time. But in the end, Dupuy’s model is not all that useful for understanding or predicting counterinsurgency. For a war in which the size of the enemy is extremely hard to discern, and the enemy’s ability to regenerate new forces through recruiting is important, such equations are difficult to use productively since they do not offer insight into such matters. The actual nature of the counterinsurgency combat that ensued (and that continues as of this writing) requires a different type of analytical approach, developed in a subsequent section in the following pages.

AMPHIBIOUS ASSAULT

The categorizations used earlier—heavy armored combat, infantry combat—are cleaner and neater than warfare often is itself. Many conflicts have elements of both these types of combat. And there are other categories of major military engagement as well. Amphibious warfare is an important such example. Its heyday was undoubtedly World War II, in the Pacific and Normandy and in Italy and elsewhere, though there were also important examples of amphibious operations at Gallipoli, Turkey in World War I, and in Inchon, Korea, and more recently in the Falklands War. The following framework is designed to help gauge the basic feasibility of an amphibious assault operation for a given military balance (not to estimate casualties, which might be best forecast by looking at the individual elements of the battle separately, with tools developed earlier in this chapter). As such, it implicitly provides a way of trying to size a force for an amphibious assault in terms of force planning.⁶⁴

The modern era of precision strike has made some doubt whether amphibious assault really has a future, and the U.S. military’s decision not to try to fight ashore in Kuwait in 1991 (largely out of concern over simple Iraqi mines at sea) reinforced the theory that perhaps amphibious operations had become too difficult in current times. But the U.S. Marine Corps, among other organizations and militaries, still takes the amphibious mission quite seriously (maintaining a sealift capacity to put nearly a division ashore in the face of hostile fire, and purchasing amphibious vehicles and tilt-rotor aircraft designed largely, if not primarily, for such operations). There may also be situations where amphibious assault is contemplated for the simple reason that it may be the only way to take a given objective.

How to assess the prospects for successful amphibious assault, as well as the likely course of battle and key determinants of outcomes? Answering these questions requires additional tools beyond those discussed previously for land warfare (though amphibious assault, if at least initially successful, ultimately morphs into land warfare, meaning that equations like those developed earlier can be used to analyze part of the problem). In fact, an amphibious attack is more complex, involving several phases of battle, and the dynamics of aerial attack, naval and air troop movements, and defensive efforts to prevent those movements complicate the analytical exercise. As such, what is needed (prior to using any mathematical formula) is a systematic way of thinking through the respective phases of an operation and asking at each stage if the attacker would possess what would likely be needed to have a chance of success at each stage.

To succeed, an invader has generally first needed air superiority, and preferably outright air dominance or supremacy. Second, the attacker has used a combination of maneuver, surprise, and brute strength to land troops in a place where they locally outnumber defenders in troops and firepower. As the previous discussion about data gathered by Trevor Dupuy, the Army’s Concepts Analysis Agency, Joshua Epstein, and others shows, it is in principle possible to win battles when outnumbered. But in the case of amphibious assault, that is difficult, since the very act of getting ashore itself provides some warning to the defender—possibly depriving the attacker of some of the surprise that might be achieved more easily on land. Third, the attacker has then generally tried to strengthen its initial lodgment faster than the defender can bring additional troops and equipment to bear at the same location.

If an attacker can do most or all of these things, it has a good chance of establishing and then breaking out of an initial lodgment. As Table 2.1 shows, attackers can succeed without enjoying all three advantages. But in the cases considered here, they did not succeed without at least two of them. For a key modern scenario, an attempted PRC conquest of Taiwan, China might (or might not) gain a large edge in the air; it could not, however, satisfy the other two criteria as discussed more below.

These historical cases, and the framework used here to explain their likelihood of success or failure, probably understates the difficulty of amphibious assault in modern times. For example, the capabilities of modern sea mines outstrip those of mineclearing technologies for the most part, especially in shallow waters.

Even more to the point, the era of precision strike makes it very hard for large assets, notably ships, to approach a defender’s shores if the defender has any reasonable combination of prepared defensive positions and advance knowledge of where the attack is to occur. The U.S. Marine Corps, recognizing these developments, has in the modern era placed a premium on maneuver and speed rather than traditional frontal attack—and as noted, the United States chose not to employ an amphibious assault against Iraq in the liberation of Kuwait in 1991.⁶⁵

TABLE 2.1
Key Elements in Amphibious Assaults

Case/Attacker	Air Superiority	Initial Troop Advantage at Point of Attack	Buildup Advantage at Point of Attack
Historical Successes
Okinawa, 1944/U.S.	yes	yes	yes
Normandy, 1944/U.S., allies	yes	yes	yes
Inchon, Korea, 1950/U.S.	yes	yes	yes
Falklands, 1982/UK	no	yes	yes*
Failed Attempts Anzio, 1943/U.S. and UK*	yes	yes	no
Gallipoli, 1915/UK, allies	no	yes	no
Bay of Pigs, 1961/Cubans	no	marginal	no

* British forces were outnumbered on East Falkland Island, but they managed to build up their lodgment successfully and move out from it without opposition. At Anzio, although the forces there ultimately contributed to allied victory in Italy in the spring of 1944, their initial objective of making a quick and decisive difference in the war during the previous winter was clearly not met; thus, the operation is classified here as a failure.

 

So how can we ascertain whether, in a possible future war, a given attacker could establish at least two, and better yet three, of the preceding criteria likely to correlate with a successful invasion? The questions on local force advantage are largely a function of the relative logistical capabilities of the attacker and defender to move forces in the face of hostile fire. The first question is slightly more complex.⁶⁶ It is not just a matter of who flies the better airplanes or the larger air force, but what effects one can achieve from the skies—and prevent the adversary from itself carrying out. Thus there is some ambiguity in determining if this criterion for possible success is met.

It is easiest to think through these complex issues with reference to a specific example, and given the importance of this scenario, a good case for scrutiny is that of a possible Chinese attack on Taiwan. As is well known, China rejects the idea of Taiwan’s permanent separation from the mainland, whereas Taiwan has been seeking to expand its role as a sovereign entity in global affairs, and many Taiwanese feel that the island should be independent. The potential exists for a political crisis that could lead to war—and a war that might ultimately even engage the United States. Even if unlikely, such a conflict is important to analyze given the huge stakes.

Chinese Attempts to Establish Air Superiority

Begin with the matter of trying to establish air superiority (this is the less demanding of the two standards of aerial advantage listed earlier, with air supremacy suggesting a much greater degree of dominance). From China’s point of view, ideally it would ground Taiwan’s air force early in any war. Otherwise, even the occasional Taiwanese attack aircraft could cause serious losses to PRC ships approaching Taiwan’s shores in an amphibious assault, given the accuracy and lethality of modern aerial ordnance, as discussed further in the following. At a minimum, China would need a major edge in the air to make such Taiwanese attacks rare and dangerous for the pilots carrying them out.

China does not have enough of an inherent advantage in air combat capability to achieve such air superiority through dogfighting alone. It would surely need to begin any war with a highly effective surprise strike against Taiwan’s air force, which could otherwise take shelter in its many hardened sites on its various airfields, and sneak out for the occasional strike mission against large PRC ships. (It is important to note that if it employed this surprise tactic, China could not start loading and sailing most of its ships towards Taiwan until after the missile and air strikes began, for fear of tipping off Taiwanese and U.S. intelligence about its intentions. In fact, the PRC would do extremely well simply to prepare its air and missile forces for the attack without having those preparations noticed.)

China’s ability to carry out such a surprise attack would depend on its missiles and aircraft based near Taiwan, primarily. China has a large ballistic-missile force and a large air force that could be used, among other missions, to attack airfields and planes on those airfields. The missiles are numerous, now totaling about 1,000 in southeastern China near Taiwan. While China’s ballistic missiles (as well as its cruise missiles) have been rather inaccurate to date, that is changing.⁶⁷ As for planes, China has about 100 airports of all kinds within 600 miles of Taiwan. Approximately 750 military aircraft are normally located at the twenty such airports used by the military.⁶⁸

Taiwan has hardened shelters for most of its fighters. Of course, it would have to keep aircraft in those shelters routinely to survive any surprise attack, or have advance warning of the attack.

Taiwan has another challenge here, too: keeping runways operational. Two to three dozen planes might be needed to shut down a given runway, or a somewhat lesser number in combination with China’s more accurate missiles.⁶⁹ Taiwan could begin to repair runways after any Chinese strike, assuming it has sufficient runway-repair equipment (and sufficiently hardened maintenance facilities and fuel distribution infrastructure).⁷⁰ China could undertake subsequent attack sorties, of course. However, Taiwan’s anti-aircraft artillery and SAMs would then be on a high state of vigilance, and the Chinese air force might well lose 5 to 10 percent of their planes on each subsequent sortie, even if able to use standoff precision-guided munitions that allowed them to stay out of the immediate environs of the airfields.

So Taiwan would have a good chance to keep large numbers of aircraft functional after a PRC surprise strike, assuming it was vigilant and careful in day-to-day security operations as well as in its hardening of key facilities.⁷¹ Given the number of shelters it owns, unless it was completely careless, Taiwan should retain about half its air force (perhaps 300 planes) after even a masterful PRC surprise strike.

The Initial Amphibious Assault

The next stage in the battle, amphibious assault, is somewhat simpler to model. It is largely a question of how fast transport ships can be loaded, sailed, and unloaded, as well as of how fast a defender’s force can be mobilized and sent to fighting positions.

While the math is simple enough to follow in general, focusing on the China–Taiwan case provides some additional clarity as an example. China has the capacity to transport 10,000 to 15,000 troops with some heavy armor by amphibious lift.⁷²

Assume for the moment that China could deploy all of these ships to a single point on Taiwan’s shores at once. The question is, could the number of arriving troops rival the number of Taiwanese defenders that would likely be able to arrive at a comparable time?

The defender’s response is a function of how soon it would notice the impending assault, how long it would then take to mobilize and activate troops, and how long it would need to position those forces where they were needed. Taiwan would have numerous advantages, starting with the fact that its own reconnaissance capabilities and those of the United States would surely pick up Chinese preparations for the assault well before ships left PRC ports. In fact, by the assumptions of this scenario, it would have been struck with a surprise attack against its airfields, providing obvious warning. The only truly plausible reason that it might not respond promptly has to do with the vulnerability of its command and control network; if key technical systems were knocked out, and if some top political and military leaders were assassinated or kidnapped, Taiwan’s response could be delayed. This underscores the importance of redundant command and control as well as clear procedures for maintaining the chain of command when it is under attack. Such matters are hard to model with simple arithmetic; they require more complex and sometimes tedious technical assessments of specific technical vulnerabilities.

Assuming that Taiwan’s command and control and communications systems largely survived the opening attack, it would then be rather well positioned to respond. With a large military and a relatively small land mass to defend, as well as a good road network, it would have advantages that not every such defender would enjoy.

To be more concrete, note that Taiwan has:

•   200,000 active-duty ground troops
•   1.5 million more ground-force reservists
•   A coastal perimeter of about 1,500 kilometers

Portioning these out, this means Taiwan could deploy roughly 1,000 defenders per kilometer of coastline along all of its shores if it wished. So over any given stretch of ten to fifteen kilometers, a tactically relevant distance (since forces over that distance could all reach incoming Chinese assets with many of their weapons), fully mobilized Taiwanese defense force would be able to deploy as many troops as China could deploy there with all of its amphibious fleet. (An attacker would need to seize a shoreline of roughly that length, to create areas safe from enemy artillery.)⁷³ This clearly assumes, however, that Taiwan would make rapid use of its reservists.

The preceding presupposes rapid Taiwanese mobilization, but no advance knowledge by Taiwan about where the PRC intended to come ashore. In reality, unless completely blinded and paralyzed by China’s preemptive attacks against airfields, ships, shore-based radars, other monitoring assets, and command centers, Taiwan would see where ships sailed and be able to react with at least some notice. (It is also very likely that, even if it did not immediately send combat forces, the United States would be willing to provide Taiwan with satellite or aircraft intelligence on the concentration of China’s attack effort. The United States and Taiwan now have a military hotline, allowing for the possibility that the U.S. global surveillance system could plug holes in Taiwan’s own capabilities or replace them after a PRC attack.)⁷⁴ Although the Strait is typically only 100 miles wide, Taiwan itself is about 300 miles long, so ships traveling 20 knots would need more than half a day to sail its full length, and could not credibly threaten all parts of the island at once. In addition, amphibious assault troops cannot come ashore just anywhere. Only about 20 percent of the world’s coastlines are considered suitable for amphibious assault. On Taiwan’s shores, the percentage is even less, given the prevalence of mud flats on the west coast and cliffs on the east.

As a practical matter, then, Taiwan would not need to mobilize all of its reservists to achieve force parity in places most likely to suffer the initial PRC attack. If it could mobilize even 20 percent of its reservists in the days that China would require to assemble and load its amphibious armada and then cross the Strait, it could achieve force parity with China along key beachlines. Thus, China would be unlikely to establish even a local temporary advantage along the section of beach where it elected to try coming ashore.

Taiwan also has at least two airborne brigades that it could use to react rapidly where China attacked.⁷⁵ They would allow it to counter China’s airlift capabilities (estimated at two to three brigades of paratroopers). China may already be adding a capacity to carry several thousand more light forces with the purchase of thirty Il-76 Russian transport aircraft.⁷⁶ However, PRC paratroopers (or troop-carrying helicopters) over Taiwan would be at great risk from Taiwanese fighters, surface-to-air missiles, and anti-aircraft artillery. Paratroopers in fixed-wing transports are particularly vulnerable in situations in which the attacking force does not completely dominate the skies and in which the defender has good ground-based air defenses.⁷⁷

Thus, China does not possess the ability to generate the second element of most successful amphibious attacks as shown in Table 2.1. Its maximum likely deployment of initial forces would at best be comparable to Taiwan’s activation of defensive forces in the same landing zone.

Moreover, the preceding analysis does not even include expected attrition to Chinese incoming forces. In reality, such losses would be enormous, both in the initial assault and in subsequent reinforcement operations. Many of the troops crossing the Strait in China’s amphibious ships would never make it to land.

Retired Captain Wayne Hughes, Jr. provides a very simple and useful algorithm for understanding why exposed ships at sea are highly vulnerable. As he points out, given the lethality of modern antiship missiles (discussed further in the following), ship defenses are best viewed as filters—taking out a certain percentage of incoming threats—rather than reliable protectors. This is true even before the point where defenses might be saturated with more incoming threats than they are even theoretically capable of simultaneously tracking and engaging. In addition, attackers can often concentrate their fire in salvos of several shots at a time, further complicating the defense’s job. Moreover, given modern homing missiles, somewhat less skill is required in firing them effectively than was the case for ordnance in many previous eras of naval combat, making for a high expected accuracy per shot. Also, modern ships are often incapacitated after just a couple shots.

So if, say, eight shots are fired at a given ship, and six are correctly aimed towards the target, only one or two might be intercepted. That means four or five might strike their targets, and often two to four would be enough to incapacitate a given ship—so in this kind of example, the attacked ship would almost surely be sunk.

More generally, Hughes’ simple “salvo equation” can be written in simplified form as follows. Assume there are two fleets, with A the number of ships in the first and B the number in the second. B(0) is the initial number of ships for B, and B(t) is the number at a subsequent time, t, after it has taken losses.⁷⁸ Those losses are:

B(0) –B(t) = [(aA –bB)/s]

Here, a is the number of accurate shots fired by A per ship, b is the number of missiles intercepted by B’s defenses per ship, and s is the survivability of the typical ship in B’s fleet. Put differently, this equation simply says that B’s losses are equal to the total number of good shots fired by A, minus the number intercepted by B, then divided by the number of shots typically required to put any given ship out of commission. So if A has five ships each shooting four accurate missiles, for a total of twenty incoming missiles, and B has five ships each capable of intercepting two missiles on average (for a total of ten intercepts), and two hits are needed to sink the typical ship in B’s fleet, then five ships will be sunk if the missile shots are well distributed:

5 = [(20 –10)/2]

The arithmetic here is clearly very simple, but does give some sense of how ships in modern battle actually perform, and what criteria determine their survival or their demise in battle. Among other things, it shows the importance of early detection (to get in the first salvo), of networking offensively when possible (to ensure an effective distribution of missile firings—not too many and not too few at any enemy ship), and of networking defensively when possible (in the event that networked ships can defend more effectively when working together; if this is not the case, the ships are often better off operating in a more dispersed manner).

Of particular relevance for this case, the formulas can easily be modified to show the effects of attacks on an approaching fleet by a land-based force. The product of the two terms aA can simply be interpreted as the number of successful shots at fleet B, wherever and whatever their origin might be.

Historically in the modern era, more than 90 percent of missiles fired at undefended ships reached their targets (with fifty-four ships sunk or otherwise put out of action with just sixty-three missiles fired). About 68 percent of missiles fired at ships that had defenses but failed to use them properly reached their targets (with nineteen ships sunk or put out of action using thirty-eight missiles). Against ships employing their defenses, about 26 percent of missiles fired reached their mark, with twenty-nine ships incapacitated in one way or another by a total of 121 missiles fired.⁷⁹ The data for these cases come from battles before the turn of the twenty-first century. Of course, there is variation from case to case but the overall trends are telling about the difficulty of defense and survivability at sea regardless.

As another way of getting a very rough quantitative grip on the problem, for the Taiwan case or another possible example, consider that the British lost six ships to missiles and aircraft and had up to another dozen damaged, out of a 100-ship task force, in the Falklands War—and that they did not generally have to approach any closer than 400 miles from the Argentine mainland during the conflict.⁸⁰ That amounts to an effective attrition rate of roughly 10 to 15 percent during blue-water operations—against an outclassed Argentine military that only owned about 250 aircraft.⁸¹ PRC losses would surely be greater against a foe whose airfields they would have to approach directly, whose air forces would likely retain at least 300 planes even after a highly effective Chinese preemptive attack against airfields, and whose antiship missile capabilities substantially exceed Argentina’s in 1982. Taiwan possesses significant numbers of anti-ship missiles such as Harpoon and its own Hsiung Feng. Weaknesses are evident in Taiwan’s capabilities to resist invasion: its air force has focused primarily on air-to-air attack, not antiship operations, and the United States has resisted providing Taiwan with certain attack capabilities out of fear they might be used provocatively. But despite these limitations, Taiwan’s panoply of capabilities is considerable, and would be potent even at night or in bad weather. The central point is that, with presumed help from American reconnaissance, and with the advantage of seeing the PRC ships coming and being able to fire at them from relatively safe positions on the shore or from land-based aircraft, Taiwan would have the advantages that the salvo equations show to be so important.

All told, the PRC would likely lose at least 10 percent of its forces just in approaching Taiwan’s coasts and fighting ashore each time it attempted another trip. This means that after five trips it could be down to 60 percent (or less) of its initial fleet and after ten trips it could be down to 35 percent.

The situation would get no better over time for China even if it could somehow establish an initial toehold on the island. As the battle wore on, Taiwan’s internal lines of communication would help even more, and its ability to reinforce its defensive position at the chosen point of PRC attack would improve further in relative terms, even as China tried to reinforce its attacking legions. By my estimates for this case, China could average deploying no more than 10,000 more troops per day to Taiwan by sea and air combined, in days 3 through 10 of the operation. This assumes that the average ship can do a round trip to Taiwan every other day, on average, including time for loading and unloading. And for days 11 through 20, the average daily flow would be only half that due to ship attrition.⁸²

The preceding assumes that China would not be able to seize and protect a Taiwanese port, but that it would instead have to rely on amphibious shipping to an undeveloped area. Taiwan would presumably mount strong defenses near major ports and airfields (and also have the capacity, if truly necessary, to destroy most of the supporting infrastructure to deprive China of the ability to employ it).

By contrast, Taiwan could on average mobilize and deploy another 50,000 troops a day to the location where China was seeking to establish a firm toehold. And its means to continually reinforce would not be nearly as vulnerable to interdiction as would China’s.

What if the PRC used chemical weapons in this part of its attack? If it could fire chemical munitions from its ship-based guns, it might be able to deliver enough ordnance to cover a battlefield several kilometers on a dimension within several minutes. China would presumably want to use a nonpersistent agent, like sarin, so its troops could occupy the area within a short time without having to wear protective gear. The effects of the weapons on Taiwan’s defenders would depend heavily on whether they had gas masks handy, the accuracy of Chinese naval gunfire, weather conditions, and the speed with which Taiwan could threaten the PRC ships doing the damage.⁸³ Historical experiences with chemical weapons suggest that China should not expect these weapons to radically change the course of battle in any event. Even in World War I, when protective gear was rudimentary, chemical weapons caused less than 10 percent of all deaths; in the Iran–Iraq War, the figure has been estimated at less than 5 percent.⁸⁴ China would need to worry that, if its timing and delivery were not good, its own mobile and exposed troops could suffer larger numbers of casualties than the dug-in defenders.⁸⁵ Using chemical weapons could also invite Taiwanese retaliation in kind against China’s relatively concentrated and exposed forces on and near the island.⁸⁶ All told, this approach would improve China’s odds of getting an initial foothold on Taiwan slightly. However, it would not change the fact that Taiwan could build up reinforcements far faster than the PRC subsequently.

Some have raised the possibility that the PRC could use its fishing fleet to put tens if not hundreds of thousands of troops quickly ashore on Taiwan. There are several important reasons not to take this threat particularly seriously, however. First, the ships could not carry many landing craft or much armored equipment. Second, Taiwanese shore-based coastal defense guns and artillery, as well as Taiwanese aircraft, small coastal patrol craft, and mines (not just advanced antiship missiles), might well make mincemeat of many of the unarmored ships, which would have to approach very close to shore in order for the disembarking soldiers not to subsequently drown.⁸⁷ Third, given the distances involved, it would be impossible to coordinate the assault very well; the ships would inevitably arrive on Taiwan’s shores in ragged staggered formations that would deny PRC troops the benefits of massed attack.⁸⁸

Extracting broader methodological lessons from this case, it is important to break down an attempted amphibious assault into several stages. And in addition to comparing broad force capabilities, such as airpower and tonnage of shipping, it is necessary to evaluate factors like the respective intelligence and monitoring capabilities of each side, the quality of air defenses, the likely care with which command and control facilities and airfields and the like have been prepared against possible surprise attack, and internal lines of communication/transportation for the defender. Being completely confident about the analytical results is difficult, since an ill-prepared defender may be shocked into paralysis or confusion by a sufficiently well-conceived surprise strike. But if analysis shows that one side would effectively need everything to go right even to have a chance at success, and if the other side is capable of maintaining a certain vigilance in its preparations and its day-to-day watchfulness, the defender can probably succeed, given the inherent difficulty of amphibious assault in the missile age.

Some of the preceding challenges facing any amphibious attacker China could address. It could build a great deal more amphibious shipping. It could, and will, continue to improve the accuracy of its missile force and its air-delivered munitions to improve capacities against Taiwan’s air force in particular. It could continue to learn more about how to disrupt Taiwan’s command and control networks. So Taiwan’s defensible position is not guaranteed to endure. It will need to continually harden key assets and devise backup command and control measures, improve runway repair capabilities as well as the strength and number of aircraft shelters, strengthen antishipping missile capabilities, and the like. Nevertheless, its current military position against amphibious assault appears quite robust.

COERCIVE USES OF FORCE: BLOCKADES
AND BARRAGES

Obviously, not every use of military force amounts to all-out war. Sometimes countries conduct limited attacks for limited purposes. Two cases of modern relevance are considered here: a possible North Korean barrage of Seoul, primarily with artillery, and a Chinese blockade of Taiwan that might or might not be coupled with missile strikes. Of course, analyzing the likely course of any such scenario is a complex matter of political-military analysis; the goal here is simply to explain some of the technical issues involved in assessing how potent any such attacks might be. These two cases dramatize the types of scenarios because Seoul is unusually vulnerable to artillery barrage (being so close to heavily armed North Korea) and Taiwan is particularly vulnerable to missile strikes or even a partially effective, “leaky” naval blockade (being so dependent on shipborne commerce and so close to China).

A North Korean Artillery Barrage of South Korea

What could North Korea do to Seoul and environs by way of bombardment even if it could not plausibly seize the South through outright invasion? If its capabilities were sufficient, South Korea might in theory be intimidated into surrender—or at least into appeasement prior to hostilities, knowing what would happen if war began (and fearful that North Korea might be willing to run the risks, given its extremist regime).

North Korea could, to be sure, seriously harm the South Korean people and economy. About 500 of North Korea’s roughly 12,000 artillery tubes are within range of Seoul in their current positions. Most artillery can fire several rounds a minute. Also, the initial speeds of fired shells are generally around half a kilometer per second. That means that even if an ROK (Republic of Korea) counterartillery radar some ten kilometers away picked up a North Korean round and established a track on it within seconds, a counterstrike would not be able to silence the off ending DPRK (Democratic People’s Republic of Korea) tube for at least a minute (and probably closer to two minutes). On average, such a tube could therefore probably fire two to five rounds, and quite possibly a dozen or more, before being neutralized or forced to retreat fully into its shelter. Some tubes may even be able to fire from protected positions, permitting them to keep up the barrage until they suffer either a near-direct hit by an artillery round or an attack from a laser-guided or satellite-guided bomb.

That means that at least several thousand rounds could detonate in Seoul no matter how hard the allies tried to prevent or stop the attack. The lethal radius of a typical artillery shell is usually thirty to fifty meters (for standard 81 mm and 155 mm rounds in the U.S. arsenal, with anywhere from seven to twenty-five pounds of explosive). That reference point, as well as historical precedents from conflicts such as the Bosnia war of 1992–1995, suggest that an average round could cause up to tens of casualties and considerable physical destruction. The end result could be up to tens of thousands of civilians wounded, with perhaps one fifth the total number dead. Attacks against Seoul would probably be much worse if they involved chemical weapons.⁸⁹

A Chinese Missile Barrage and Blockade of Taiwan

Two of the most notable ways that China might threaten Taiwan, short of attempting a very risky amphibious assault, are missile attack and naval blockade. The two techniques might also be combined in a single operation.

China has about 1,000 short-range missiles, believed to be equipped with conventional warheads, in the southeastern part of its country near the Taiwan Strait. The missiles could be used in a number of ways, going well beyond what happened in the mid-1990s (when they were fired into the sea near Taiwan). They could be aimed at remote farmland or mountains, to minimize the risk of casualties (if only a few missiles were used, in a strictly symbolic way against such sites, it is plausible no one would be killed, though China could not be sure in advance). They could be aimed very close to land so they would be visible to residents when their warheads exploded. They could also be aimed at the waters just outside ports, or even within harbors, implicitly threatening Taiwan’s economy but again without being likely to cause many casualties. If the crisis intensified, successive missile strikes might be aimed closer to shore and closer to cities, with a greater risk of casualties—potentially causing dozens of fatalities in a single strike. Missiles could also be directed at military installations, if China wished to avoid civilian casualties.

The missile option has limits, however. Most of China’s ballistic missiles are not yet very accurate, with expected miss distances of 100 meters or more.⁹⁰ The types of strikes mentioned here, while perhaps unlikely to cause many casualties, would also be unlikely to achieve much direct and lasting military or even economic effect. And escalation would be problematic for China. Using missiles against cities would be seen as a brutal terroristic act that could do more to unify the people of Taiwan—and the world—against China than to achieve Beijing’s war aims. At least, that has been the historical norm when cities have been bombarded in the modern era, be it by airplanes in World War II or ballistic missiles in the Iraq–Iran and Persian Gulf wars.⁹¹ For these reasons, missile strikes might be a logical way for China to begin any use of force, but it would probably need a backup option in case they failed.⁹²

Cruise missiles can be much more accurate, and China is obtaining these, too. It may have several hundred with sufficient range to find targets in Taiwan. The warheads on these missiles may be smaller, and their likelihood of being shot down much greater, but this threat may be, on balance, somewhat more militarily meaningful. Still, perspective is needed; the United States has frequently used cruise missiles in modern war, and often used several hundred in a given conflict, but has never come close to achieving widescale military objectives with such missiles alone. In modern wars, it has typically had to deliver many thousands of precision bombs to achieve its goals.⁹³

That is where a naval blockade could offer appeal to China. It could be “leaky” and still directly threaten Taiwan’s economy. To do so, it need not physically stop all ship voyages into and out of Taiwan. It would simply need to deter enough ships from risking the journey that Taiwan’s economy would suffer badly. The goal would be to squeeze the island economically to a point of capitulation. This solution could seem quite elegant from Beijing’s point of view—it could involve only a modest loss of life, little or no damage to Taiwan itself, more terror than harm suffered by the people of Taiwan, and the ability to back off the attack if the United States seemed ready to intervene or if the world community slapped major trade sanctions on China in response. Additionally or alternatively, the capabilities needed to carry it out, most notably submarines as argued in the following, could also help deter and complicate any American naval intervention on behalf of Taiwan.⁹⁴ How to analyze such a blockade systematically and quantitatively, when its principal goals might be qualitative and psychological?

A Chinese blockade could take a number of forms. Militarily speaking, the least risky and most natural approach would simply attempt to introduce a significant risk factor into all maritime voyages in and out of Taiwan by occasionally sinking a cargo ship with submarines or with mines China laid in Taiwan’s harbors.⁹⁵

Using airplanes and surface ships would put more of its own forces at risk, especially since it could not realistically hope to eliminate Taipei’s air force with a preemptive attack (though airpower might be used in a hit-and-run raid, especially as an initial strike before Taiwan’s defenses were fully alerted). A blockade using planes and surface ships would also be rather straightforward for the United States to defeat quickly. It benefits from superior scouting/reconnaissance abilities at sea; in addition, the lethality of modern naval weapons means that the side able to muster an effective attack first is increasingly in the dominant position.⁹⁶

To be sure, a blockade would be challenging and dangerous for China to pull off. Perhaps the greatest worry for Beijing would be its likely inability to distinguish one country’s merchant ship from another’s. But if Beijing announced to the world that those shipping towards Taiwan were aiding and abetting its enemy, and gave fair warning, it might consider itself to have done enough to warrant attack against any vessel not heeding its demands. Moreover, it might offer countries the option of first docking in a PRC port for inspection (if it decided to allow humanitarian goods through, for example, or ships from certain friendly countries but not others) and then being escorted safely to Taiwan. Since this strategy might require it to sink only a few ships to achieve the desired aims, even in a worst-case scenario it might believe it was threatening the lives of only 100 to 200 commercial seamen. Given the perceived stakes involved, Beijing could well consider this a reasonable risk.

Most of China’s submarines do not have antiship cruise missiles or great underwater endurance. However, the PRC submarine force is steadily improving. Even today, Chinese subs have adequate ranges on a single tank of fuel—typically almost 10,000 miles—to stay deployed east of Taiwan for substantial periods. Although their ability to coordinate with each other and reconnaissance aircraft is limited, that might not matter greatly for the purposes of a “leaky” blockade. Even if picking up commercial ships individually by sonar or by sight, such submarines could maintain patrols over a large fraction of the sea approaches to Taiwan. It could take Taiwan weeks to find the better PRC submarines, particularly if China used them in hit-and-run modes.⁹⁷ Given the lethality of modern torpedoes and cruise missiles (for any PRC submarines carrying the latter), the existence of these submarines in important waterways near Taiwan would constitute a very major threat.

To break the blockade, the basic idea for the United States and Taiwan would probably be to deploy enough forces to the Western Pacific to credibly threaten the following type of operation. The magnitude of this operation shows how hard it is to reliably defeat even an imperfect blockade.

To break the blockade, the United States and Taiwan would have to set up a safe shipping lane east of Taiwan. In addition, they would have to heavily protect ships during the most dangerous part of their journeys when they were near the island. To carry that mission out, the United States, together with Taiwan, would need to establish air superiority throughout a large part of the region, protect ships against Chinese submarine attack, and cope with the threat of mines near Taiwan’s ports.

The anti-submarine warfare (ASW) effort could have multiple aspects. The United States would surely be tempted to deploy its own attack submarines as close as possible to China—certainly in the Taiwan Strait, maybe just outside PRC ports. This approach would provide American submarines a good prospect of destroying PRC subs at their source, before they were in a position to fire on commercial shipping (or U.S. aircraft carriers) in more distant waters. However, this type of ASW would be extremely delicate strategically, especially if it involved attacks in Chinese territorial waters.

Whatever happened near Chinese shores, there would surely be additional layers of American ASW further out to sea. The convoys sailing to and from Taiwan would need protection. American ships, primarily ASW frigates, would accompany convoys of merchant ships as they sailed in from the open ocean waters east of Taiwan. These convoys might form a thousand miles or more east of Taiwan, and enjoy armed protection from that point onward as they traveled to the island and later as they departed. The frigates would use sonar to listen for approaching submarines, and for the sound of any torpedoes being fired. Some ships would be larger destroyers or cruisers, such as those equipped with advanced Aegis radars, to detect any use of cruise missiles and attempt to defend against them.

The United States would probably deploy significant numbers of surface combatants and airplanes like P-3s to the region for this mission. Some would help protect U.S. aircraft carriers, of which at least four would likely deploy east of Taiwan to establish air superiority in the event of any conflict. Others would provide additional protection to merchant ships or mine warfare vessels as they operated near Taiwan’s shores. U.S. minehunters and minesweepers would operate near Taiwan’s ports and the main approaches to those ports. Land-based or ship-based helicopters might assist them. So might robotic submersibles deployed from ships near shore.

If China then used its submarines in attacks on shipping, or if direct hostilities began in another way, the United States would almost surely begin to actively search for and fire upon Chinese submarines as a matter of normal operations. Any Chinese submarine wishing to fire at a merchant ship or aircraft carrier would then first have to run quite a gamut. It would have to evade submarine detection as it left port, avoid any open-water search missions that the United States and Taiwan established, and then somehow penetrate the defensive ASW perimeter of whatever convoy it was attacking as it approached its target. To survive the overall engagement and return to port, it would then need to successfully negotiate all of this in the other direction.

During the Cold War, the effectiveness of ASW operations was commonly assessed at 5 to 15 percent per “barrier” (Cold War barriers at that time were more linear and literal perimeters than would be likely here, but the fact remains that Chinese subs would have to survive perhaps three types of pursuers at three different parts of their journey to or from home base.) By those odds, the typical Chinese sub would do well to survive for two or three roundtrip missions from base.⁹⁸ But it might succeed in getting off several shots against valuable surface ships before meeting its own demise.

The vulnerabilities of all ships (including U.S. Navy ships) to attack are amplified in shallow littoral waters. In the open blue waters of the oceans, the U.S. Navy can generally detect enemy ships or aircraft long before they are close enough to strike. But in shallow waters, shore-launched antiship missiles are a threat, as are weapons fired from aircraft or ships that dart out from a country’s coastal regions. A similar conclusion applies to the threat from submarines and torpedoes. In the open oceans, the U.S. military can rely on sonar (from aircraft, fixed underwater arrays known as SOSUS, and ships) to get a good sense of the approach of enemy submarines. Sonar is relatively predictable in deep waters; moreover, any ship approaching a U.S. vessel would have to travel a great distance to reach it in such a location, offering multiple opportunities for detection. By contrast, shallow waters are complex sonar environments, where sound waves bounce back and forth in multiple and unpredictable directions. This makes ambush a real worry, especially for the mine warfare vessels and surface ships that would have to escort commercial vessels all the way into Taiwan’s ports.⁹⁹

To be sure, the United States Navy would not deploy most of its assets near China all at once. However, China still might think that a quick strike that sank a carrier and killed hundreds or thousands of Americans would cause Washington to waver in its future commitment to the defense of Taiwan. As naval analyst Captain Wayne Hughes, Jr. argues about the nature of naval combat:¹⁰⁰

•   Defense is usually weaker [than offense, unlike land battle].
•   Defensive power is solely to gain tactical time for an effective attack or counterattack.
•   When two competitive forces meet in naval combat, the one that attacks effectively first will win.

Hughes also argues that there is perhaps less friction in battle at sea than on land, and that many naval engagements occur in which decisive results happen relatively quickly, in contrast to the norm for war on land.

Technology trends could put ships at even greater risk today. Hypersonic antiship cruise missiles are becoming more common and are extremely difficult to defend against, even for high-performance U.S. Navy ships with advanced Aegis radar systems. The ranges of PRC cruise missiles are now reaching or exceeding 150 miles. To make these weapons more effective, China can be expected to try to improve its targeting and communications systems, too. For example, it is putting into orbit more satellites capable of detecting large objects on the oceans.¹⁰¹ With the information from satellites, guidance systems on the cruise missiles could then guide them to the vicinity of their targets, where terminal seekers on the missiles themselves could finish the navigation job.¹⁰² The only truly reliable way to protect ships against such threats is to minimize the number of missiles that can be fired at them by depriving the missile-carrying aircraft or ships of proximity to their would-be targets. The United States would have a good chance to do this successfully against China, but the PRC’s submarines would complicate the task and cause a real risk of significant American losses in the course of battle.

NUCLEAR EXCHANGE CALCULATIONS

Although more a vestige of the Cold War than a focus of current defense planning, nuclear “exchange calculations” are still worth understanding. The physics underlying them remain relevant even if the nuclear superpower dynamics that led to their centrality in defense circles now seem (mostly) anachronistic. More to the point, perhaps, certain aspects of these calculations could be relevant in considering other issues, such as possible nuclear exchanges between regional powers or between a regional power and the United States—and perhaps most importantly, arms control arrangements designed to lower the probability of such nuclear attacks.

These exchange calculations focused on a nuclear-armed “triad” of forces that the Soviet Union and the United States maintained during much of the Cold War—and that Russia and the United States maintain today, at lower force levels. Smaller nuclear powers typically aspire to a similar distribution of nuclear forces across different types of platforms and delivery vehicles. The triad is made up of intercontinental ballistic missiles (ICBMs), historically placed in the ground in hardened concrete silos or made mobile to complicate the enemy’s targeting; submarine-launched ballistic missiles (SLBMs) on submarines that can be deployed at sea to enhance their survivability against attack; and bomber-launched nuclear munitions (bombs or cruise missiles), delivered by planes that can if necessary be placed on runway alert (or even be maintained with some fraction constantly airborne), again to enhance survivability.

An essential goal of nuclear force planners for much of the Cold War was to ensure that their own country’s forces could survive any plausible attempt by the enemy to disarm them in an all-out surprise first strike. At the same time, force planners also sought, especially on the American side early in the Cold War, to maximize their capacity for denying the adversary the very survivable second-strike force that they knew necessary for their own country. Worries about ensuring a survivable deterrent drove the two states to create triads in the first place, to harden silos (or, later, build mobile ICBMs that could not be easily pinpointed by the potential enemy), build quieter and quieter (hence harder and harder to find) submarines, put bombers on runway alert at interior bases, and build early detection radars and satellites to watch for any possible “bolt from the blue” attack by the other.¹⁰³

In assessing whether they had achieved their core defensive goal of creating survivable forces (and in also assessing whether they could destroy much of an enemy’s nuclear assets through “counterforce” strikes), nuclear planners employed exchange calculations. The basic logic went like this. Assuming one side might be willing to launch an all-out zero-warning attack on the other, how well might it do?

The details of the calculations were based on the following basic concept. First, ICBM silos would be attacked by the other side’s most accurate and lethal ballistic missile warheads, since they are difficult to destroy. Typically, two warheads would be used against each silo, to account for the imperfect reliability and accuracy of the incoming warheads. (It was further assumed that by the time one side’s bomber forces could penetrate the airspace of the other, the attacked side would have had six to ten hours of warning—allowing it to launch its own ICBMs before they could be attacked. Thus it was generally assumed that bombers would not be the appropriate delivery vehicle for attacking silos.) Submarines carrying SLBMs but located in their ports would be destroyed (a fairly straightforward proposition, since submarines are not very hard relative to ICBM silos); deployed submarines would be hunted down by the other side’s attack submarines (probably something that only the United States ever had the capacity to do). Bomber bases could be barraged, again by warheads from ballistic missiles. Any bombers that had managed to become airborne before the barrage occurred would have to be shot down by air defenses. This was the basic picture of any first strike.

Many assumptions, besides the previously noted point about bombers not being well suited to attack ICBM silos, were built into these models—and not all of them were guaranteed to be right. One assumption was that neither side would launch its ICBMs and SLBMs in the fifteen to twenty minutes it might have available, between when it detected the other side’s massive launch of missiles and when the incoming warheads would begin to detonate. A second assumption was that the command and control and communications systems of each side would survive initial attacks and be capable of ordering and coordinating retaliation; otherwise, having one’s forces survive might be of little benefit.¹⁰⁴ A third assumption behind the way nuclear force planning was done (if not the exchange calculations themselves) was that keeping forces on alert would not lead to a high risk of accident that would trump the strategic benefits.¹⁰⁵ A fourth assumption was that any conventional war that preceded the nuclear exchange would not lead to confusion about whether a nuclear attack had begun (when, in fact, it had not).¹⁰⁶ Again, if incorrect, this assumption would not invalidate the exchange calculation methodology per se, but it would call into question the logic of the prevailing force planning paradigms.

Some of these assumptions can be assigned a probability, or be analytically assessed; others are more intangible. For example, the idea that neither side would launch on warning was hardly an inevitability and may well have been wrong at many times during the nuclear age—but it still was a useful simplifying way to assess the maximum vulnerability of a given side’s forces. The assumptions about the survivability of command and control, by contrast, could in large measure be subject to detailed technical examination of the effects of nuclear blasts on various technologies integral to the command, control, and communications effort. In fact, many steps had to be taken to ensure survivable command and control since technical studies often revealed vulnerabilities—with both superpowers naturally trying to diagnose and repair their respective vulnerabilities before they were recognized by the other. The hope that accidents would not occur could be evaluated either by appeal to organization theory (trying to find analogies for how well large organizations had maintained strong safety records) or by examination of the history of the nuclear weapons business itself, once enough years had passed to create a reasonable dataset of experiences with airborne bombers, at-sea submarines, and the like.

In addition to such assumptions, calculations also were done with certain simplifications. For example, it was assumed that two warheads could be detonated near a single silo, without the shockwave, x-rays, or debris cloud from the first destroying the second (a process known as “fratricide”).¹⁰⁷ It was further assumed that missiles would perform the same way on flight trajectories over the North Pole as they had on test ranges. It was also assumed that, while ballistic missiles might individually fail, there would be no systemic problem with warheads detonating—even though those warheads would never have been tested under truly realistic conditions before.

Keeping these potentially flawed assumptions in mind, this is how the exchange calculations then proceeded for some typical weapons involved. Assume, say, Soviet SS-18 missiles (with ten warheads each) attacking American Minuteman missile silos. The reliability of any SS-18 missile itself was estimated at 85 percent. The average miss distance (or “circular error probable”) of a warhead launched by an SS-18 was estimated at 150 meters. Since the warhead yield was estimated at 500 kilotons, corresponding to a “lethal radius” of 290 meters against American ICBM silos (given their specific level of hardening, estimated at a resilience of up to 2,000 pounds per square inch of overpressure from a blast wave), that meant most SS-18 warheads would destroy the silo they were aimed at (since most would land within 290 meters of their aim points, given their typical miss distance of only half that distance). The main hope for the silo, by this methodology, would be that any and all missiles launching warheads at it would fail. Assuming two different missiles were used to launch the two warheads directed at each silo—a practice that complicates targeting, of course, but hedges against the failure of any given rocket—this implies a 95 percent chance of any given U.S. Minuteman missile being destroyed.¹⁰⁸

For submarines, the math was simple. All SLBMs in port would be destroyed; all U.S. SLBMs deployed at sea would survive. Soviet-deployed SLBMs, for their part, might or might not survive depending on how well U.S. attack submarines and other antisubmarine warfare capabilities were able to do their jobs in finding the Soviet subs carrying the SLBMs.

For bombers, all bombers on the ground would be assumed to be destroyed, all those well into the air before the attack would survive (and then stand some chance, perhaps 50 to 90 percent, of successfully penetrating enemy air defenses as they approached their targets). Only for those trying to get off the ground as warheads started to explode around them would the math be complicated, as the attacker might have sought to barrage the airspace around the runways, creating enough overpressure to knock planes only just getting off the ground out of the sky. The mathematics of this process are not discussed in detail here. Suffice it to say that a warhead of roughly 500 kilotons’ yield could likely destroy aircraft out to about three to five kilometers distance (though there is considerable uncertainty in this estimate and it could easily vary by a factor of two or even more, either way).¹⁰⁹ However, since planes travel fast and can vary their flight direction and altitude quickly, a barrage scenario is a complex one to evaluate.

Typically, exchange calculations showed that either superpower might retain 20 to 50 percent of its initial forces after an enemy first strike during most of the latter decades of the Cold War. Each would still have retained a great deal of redundancy or “overkill” in its nuclear forces, clearly. The situation clearly might not be so simple, however, for smaller nuclear weapons, and the potential for high vulnerability could be much greater.

SIZING STABILIZATION AND PEACEKEEPING FORCES:
THE CASE OF A COLLAPSING PAKISTAN

How should peacekeeping or stabilization missions be sized and structured? This is a difficult question, because there is no simple one-size-fits-all formula. In cases where parties to a conflict are truly exhausted by fighting and committed to peace, or cases where available weaponry is limited, for example, modest numbers of troops may suffice. Then again, it is this logic—or this sort of hopefulness—that may have led to the carelessness with which the United States prepared for the occupation of Iraq. In cases where the risk of failure is truly not acceptable, what would cautious defense planning metrics say about necessary troop levels?

This section of the chapter employs a concrete example to motivate and illuminate an otherwise rather dry set of calculations. The case in point is Pakistan. Of all the military scenarios that would undoubtedly involve the vital interests of the United States, short of a direct threat to its territory, a collapsed Pakistan ranks very high on the list. This is not the usual peacekeeping scenario, to be sure, and it is (one hopes) extremely unlikely to occur. But its potential importance makes it a good subject for discussion and analysis—and Pakistan’s sheer size underscores the tyranny of the arithmetic that drives this type of force planning. This is because the most notable characteristic of this type of military analysis is the direct relationship between the size of the indigenous population in question and the resulting necessary size of the stabilization force.

The combination of Islamic extremists and nuclear weapons in Pakistan is extremely worrisome. Were parts of Pakistan’s nuclear arsenal ever to fall into the wrong hands, al-Qaeda could conceivably gain access to a nuclear device with terrifying possible results. Another quite worrisome South Asia scenario could involve another Indo–Pakistani crisis leading to war between the two nuclear-armed states over Kashmir.¹¹⁰

The Pakistani collapse scenario appears unlikely given that country’s relatively pro-Western and secular officer corps.¹¹¹ But the intelligence services, which effectively created the Taliban and have also condoned, if not abetted, Islamic extremists in Kashmir, are less dependable. Plus, the country as a whole is sufficiently infiltrated by fundamentalist groups—as the attempted assassinations against former President Musharraf (and other evidence) make clear—that this terrifying scenario of chaos cannot be dismissed.¹¹²

Were it to occur, it is unclear what the United States and likeminded states would or should do. It is very unlikely that “surgical strikes” could be conducted to destroy the nuclear weapons before extremists could make a grab for them, since it is doubtful the United States would know their location, and it is at least as doubtful that any Pakistani government would countenance such a move, even under duress.

If a surgical strike, series of surgical strikes, or commando-style raids were not possible, the only option might be to try to restore order before the weapons could be taken by extremists and transferred to terrorists. The United States and other outside powers might, for example, respond to a request by the Pakistani government to help restore order. But given the embarrassment associated with requesting such outside help, it might not be made until it was almost too late. Hence, such an operation would be an extremely demanding challenge, but given the stakes, there might be little recourse than to attempt it.

What could this type of mission entail? The international community might team with Pakistani forces to help defeat any insurrection. Or it might help protect Pakistan’s borders, making it hard to sneak nuclear weapons out of the country, while providing only technical support to the Pakistani armed forces as they tried to put down the insurrection. All that is sure is that, given the enormous stakes, the United States would literally have to do anything it could to prevent nuclear weapons from getting into the wrong hands.

Should stabilization efforts be required, the scale of the undertaking could be breathtaking. Pakistan is a very large country. Its population is about 170 million, or six times Iraq’s. Its land area is roughly twice that of Iraq; its perimeter is about 50 percent longer in total. Since a U.S. force of some 140,000 (as part of a broader international force of about 165,000) was ultimately determined to be inadequate to handle the Iraq mission, stabilizing a country of Pakistan’s size might seem to require more than a million foreign troops.

To be more precise, according to optimal counterinsurgency doctrine, stabilizing a given region should ideally involve twenty to twenty-five military or police personnel for every thousand indigenous citizens.¹¹³ Put differently, that is at least one peacekeeper for every fifty citizens, implying in Pakistan’s case up to three million security force personnel. Such ratios have sometimes been achieved, notably in post–World War II Germany, in Bosnia and Kosovo in the 1990s, and even in Somalia in the early 1990s for a time.¹¹⁴ Even a more modest approach that accepts somewhat greater risk suggests deploying one peacekeeper for every one hundred members of the civilian population.¹¹⁵ Only in relatively benign post-conflict environments with an exhausted or thoroughly defeated population (Japan in 1945, Namibia in 1989, El Salvador in 1991, Mozambique in 1993) has it been possible to make do with only one peacekeeper for every 200 or more citizens of the host nation, still implying almost one million forces for Pakistan’s population.¹¹⁶

This arithmetic quickly shows that no international force could do the job on its own. The world does have more than 20 million soldiers across all of its militaries, but obviously most are unavailable at any given moment for peace operations and stabilization missions and most countries are unable to deploy and sustain substantial numbers of forces abroad in a timely fashion even when they choose to try. The international community has reached new heights in recent years deploying some 150,000 or more total troops in various peace operations (including those under U.N. auspices, as well as under organizations like NATO or the African Union), and a comparable number in Iraq. Its maximum capacity is no more than twice that aggregate amount, or a bit more than half a million troops, even assuming an all-out American effort. That size force could only be deployed over a period of many months—and longer still for some inland locations, given the logistical limitations of many militaries and the corresponding need to establish transportation and support for them (through contractors or other means).¹¹⁷

Presumably even in a scenario of a gradually fraying or dissolving Pakistan, some fraction of the country’s own security forces would remain intact, able, and willing to help defend their country. By this logic, the international force would deploy only at the request of the host government—which would, in any case, have to continue to handle most of the challenges of the job itself for months as the international force deployed.

Pakistan’s military numbers 550,000 Army troops, 70,000 uniformed personnel in the Air Force and Navy, another 510,000 reservists, and almost 300,000 gendarmes and Interior Ministry troops.¹¹⁸ Police forces are also substantial. But if some substantial fraction of the military broke off from the main body, say a quarter to a third, and were assisted by extremist militias, it is quite possible that the international community would need to deploy 100,000 to 200,000 troops to ensure a quick restoration of order. Given the need for rapid response, the U.S. share of this total would probably be a majority fraction, or quite possibly 50,000 to 100,000 ground forces. Obviously, this calculation is notional and illustrative, not precise; in a given scenario, the numbers could be much different.

Of course, proper force sizing does not guarantee a successful outcome. History contains numerous examples of failures even when theoretically adequate forces were deployed, as with the French experience in Algeria from 1954–1962 (about forty troops per 1,000 indigenous citizens) and the U.S. experience in Vietnam (about eighty-five troops per 1,000 Vietnamese). Clearly a host of other factors impinge on a mission’s prospects, including the training and quality of the counterinsurgency/stabilization forces, the perceived political legitimacy of their mission, and the availability of sanctuaries in neighboring states for insurgents. But being wary of the importance of force numbers is one thing; totally ignoring their relevance is something else. Without adequate forces, it is impossible to protect the population, and counterinsurgents often fall back on excessive use of firepower to compensate for their lack of presence.¹¹⁹

ASSESSING COUNTERINSURGENCY
AND STABILIZATION MISSIONS

How to tell if a counterinsurgency war is being won? Clearly sizing the force correctly for a stabilization mission, as discussed earlier, only addresses the question of getting the inputs roughly right. What about results on the ground?

In conventional warfare, deducing trends is fairly obvious—if not to predict outcomes, at least to discern who is “ahead” at a given moment. Movement of the front lines, industrial production of war matériel, and the logistical sustainability of forces in the field provide fairly obvious standards by which to assess trends. But counterinsurgency and stabilization operations—like the one in Iraq (used here as an important recent example)—are different, and more complex. How do we measure progress in such a situation?

This is a hard challenge because it is easy to misuse and abuse metrics. In Vietnam, for example, the United States was convinced that there would be a “crossover point” in attrition of the Viet Cong. If U.S. military forces could manage to kill enough of them, say 50,000 a year, their recruiting efforts would not be able to keep pace, and combined American and South Vietnamese forces would ultimately prevail. This was based in part on the conviction that successful counterinsurgency requires ten government soldiers for every insurgent—a simplifying assumption that, while partly validated by history, gave American policymakers too much confidence that a given number of U.S. troops could produce victory. That approach led to General Westmoreland’s famous search-and-destroy concepts for ground operations. It resulted in a focus on massive firepower that killed huge numbers of innocents and failed to achieve its military objective as well. The conviction that the Viet Cong needed hundreds or thousands of tons of supplies daily led to additional bombing of the Ho Chi Minh trail and ultimately Cambodia—again to no avail as it turned out that the Viet Cong in South Vietnam needed little outside help.¹²⁰

The U.S. focus on supporting a government with strong anti-communist credentials led to dependency on a corrupt regime with limited legitimacy among its people. American hopes about sparking GDP growth in Vietnam were dashed because the country’s economic successes accrued only to a small fraction of the population. Finally, Washington’s focus on enlarging and equipping South Vietnamese security forces could not compensate for their qualitative deficiencies.¹²¹

The experience of successful counterinsurgency and stabilization missions in places such as the Philippines and Malaya (now Malaysia), by contrast, tends to place a premium on tracking trends in the daily life of the typical citizen. How secure are they, and who do they credit for that security? How hopeful do they find their economic situation, regardless of the nation’s GDP or even their own personal wealth at a moment in time? Do they think their country’s politics are giving them a voice?¹²²

The Marine Corps tended to focus on these metrics in Vietnam, and developed an approach called the Combined Action Program (CAP) to help protect the population in “ink spots” that would gradually expand with time. In fact, the Marine CAP concept applied more broadly would have led to fewer overall American forces than were actually deployed, suggesting that the ten-to-one rule was NOT the optimal way to gauge U.S. force requirements. But the Marine Corps did not carry the day with this concept for the U.S. military overall.¹²³ The U.S. military finally moved towards this type of thinking in Iraq—but, in general, not until 2007.¹²⁴

However, tracking trends in the well-being of a population is extremely difficult. Many considerations enter into this question. This helps explain why the Iraq Index at Brookings has included more than fifty key indicators since we began it in late 2003. Further complicating matters is that information is often unreliable. In Iraq, data has been conspicuously poor about the unemployment rate, the crime rate, and trends in the availability of water, sewage treatment, and health care. Finally, in gauging whether a given strategy is working, some metrics may be leading indicators of success (or failure), whereas others may lag. Is there a way to make sense of the cacophony of data?

Once a counterinsurgency has made major strides forward, as Iraq had by this writing in late 2008, it is easier to tell a clear story from the data. Beginning with civilian fatalities from violence (of all forms), perhaps the ultimate indicator of stability, Iraq’s rate of violence had declined by 80 percent relative to the 2006 peak and was even lower than in the years 2004/2005. Moreover, it was continuing to decline even as the U.S. surge of forces ended and America reduced its combat brigade strength from twenty to fifteen. With U.S. troop fatality rates down by 60 to 80 percent by mid-2008 as well, and Iraqi Security Force casualties reduced by more than half, too, the overall trajectory of the war was fairly unambiguously good—just as it had been unambiguously bad in 2006.

More complicated is to assess the dynamics of such a war as the situation is changing but before trends are dramatic. This was essentially the situation in Iraq in 2007. As the surge brigades arrived from the United States over the first half of the year, and Iraqi security forces continued to grow at a rate of almost 10,000 uniformed personnel a month, new operations were initiated and the battlefield changed substantially. Nonetheless, violence remained very high; it was not until the latter half of the year that the situation markedly improved throughout much of the country. The U.S. political debate over the surge was meanwhile quite acute, and Congress was considering cutting off funding for the war even as the surge began during the first six to eight months of the year. What indicators could it have looked to, during this transitional time, to determine whether it was worth keeping American forces involved in the fight?

It is difficult to create a clearly prioritized list because leading and lagging indicators could vary from one conflict to another. In Iraq, reductions in U.S. and Iraqi security force casualties lagged because the surge led to heavy fighting in parts of the country as Shiite militias, al-Qaeda in Iraq extremists, and others battled back for a time. Improvements in basic economic-quality-of-life indicators, such as numbers of children in school, the quality of health care, the unemployment rate, and the availability of potable water and electricity continued to lag even in late 2008.

In other cases, the nature of the information may be inherently difficult to interpret. For example, “body counts” of killed enemy combatants may indicate progress—as long as the right people are being killed. But if innocents, or would-be allies, are killed by government forces, the effect can be negative. The latter dynamic probably existed in Iraq in 2004 through 2006; the former appears to have been established by 2007, but body counts themselves would not show the change.

By contrast, some indicators were more promising. Ethnic cleansing rates declined by mid-2007. The numbers of extremist leaders purged from the Iraqi Security Forces and other Iraqi government positions increased quite a bit (though it took a while to be confident that their replacements had higher integrity). Increases in the number of Iraqi security forces taking primary responsibility for local security were also encouraging. But we did not yet know for sure, in 2007, if they would be able to do so in the ethnically mixed neighborhoods in and around Baghdad, Mosul, and Kirkuk or in particularly tense regions like Basra and Sadr City. Only in the spring of 2008 were improvements in Iraqi forces validated by battlefield progress in such places.

Political progress in Iraq was slow through most of 2007, though it picked up as the year unfolded. Knowing how to gauge political progress is hard. It is not a matter of meeting specific “benchmarks” so much as creating a spirit of nonviolent politics and compromise, so that future disputes will be settled in the halls of parliament rather than on the streets or battlefields. Benchmarks are ways of gauging possible progress towards this attitude, but no more than that, and as such must be taken with grains of salt. In 2007, the main progress was in purging extremists from government jobs (as noted earlier), in Baghdad sharing more resources with Iraq’s eighteen provincial governments, and in deploying Iraqi forces to places where they could support the U.S.-led surge.

My own confidence in the new strategy grew greatly after a trip to Iraq in mid-2007, but the data were not totally conclusive at that point. It was the combination of some encouraging data trends with a general sense that the United States and Iraq had developed a proper counterinsurgency and stabilization strategy that gave me (and colleague Kenneth Pollack) confidence—underscoring again that quantitative metrics must often be married with military and strategic judgment to reach bottom-line policy judgments in this field. The science of war only goes so far.

By early 2008, things had improved much more. Progress was evident in a new pensions law, in amnesty legislation for some militia fighters, in an improved de-Baathification statute, and in a provincial powers act. Jason Campbell and I hazarded an estimate that Iraq’s politics merited a “score” of roughly 5 on a scale of 0 to 11 (using eleven benchmarks for these purposes). This was an imprecise approach, subject to future revision, but seemed the best way to gauge progress on issues that were both inherently important and topical within Iraq. Since then, progress has varied. We accorded the Iraqis 0 for resolving the logjam over the disputed city of Kirkuk’s future, for creating a permanent hydrocarbons law, and for passing a provincial election law in mid-2008, but the situation was still unmistakably improved relative to 2007 or earlier periods. By the end of 2008, our “score” for the Iraqi political system was a 7.

Other cases would require different frameworks. In fact, Iraq itself will require a dynamic approach to assessing political progress in the months and years ahead, given the nature of this business. In the counterinsurgency and nation-building business, no formula can determine whether success or failure is a certainty. But a careful and broad use of multiple metrics can help detect key trends, diagnose problems, and test various theories about whether net progress is occurring.

 

The following examples help illustrate the uses of some of these modeling and analysis methods. Some focus on current potential scenarios, while others are more generic or historical in character. Some employ a more quantitative approach than others, but all seek to use a systematic and step-by-step approach to answering the questions posed—that is what I mean when referring to combat modeling.

 

QUESTION 6: What scale of military operation might be needed to secure Kashmir or Congo or Indonesia?

 

ANSWER: Like the earlier case study on the possible collapse of Pakistan, this is a question of force planning rather than combat simulation or modeling.

Consider first a scenario pitting Pakistan against India over Kashmir. It is highly doubtful the United States would ever wish to actively take sides in such a conflict, allying with one country to defeat the other. Its interests in the matter of who controls Kashmir are not great enough to justify such intervention, and no formal alliance commitments oblige it to step in. Moreover, the military difficulty of the operation would be extreme, in light of the huge armed forces arrayed on the subcontinent and the inland location and complex topography of Kashmir. In addition to the numbers associated with Pakistan, India’s armed forces number 1.3 million active-duty troops, and feature such assets as 4,000 tanks, sixteen submarines, and about 600 combat aircraft (defense spending in the two countries was roughly $5 billion in Pakistan and $25 billion in India in 2006, respectively).¹²⁵

However, there are other ways in which foreign forces might become involved. If India and Pakistan went up to the verge of nuclear weapons use, or perhaps even crossed that threshold, they might consider what was previously unthinkable to New Delhi in particular—pleading for help to the international community. For example, they might agree to allow the international community to run Kashmir for a period of years. After local government was built up, and security services reformed, elections might then be held to determine the region’s future political affiliation, leading to an eventual end to the trusteeship. While this scenario is admittedly a highly demanding one, and also unlikely in light of India’s adamant objections to international involvement in the Kashmir issue, it is hard to dismiss such an approach out of hand if it seemed the only alternative to nuclear war on the subcontinent. Not only could such a war have horrendous human consequences—killing many tens of millions—and shatter the tradition of nuclear non-use that is so essential to global stability today. It could also lead to the collapse of Pakistan, and thus the same types of worries about that country’s nuclear weapons falling into the wrong hands discussed earlier in these pages.

What might a stabilization mission in Kashmir entail? The region is about twice the size of Bosnia in population, half the size of Iraq in population and land area. As noted in the earlier section, according to optimal counterinsurgency doctrine, stabilizing a region of 10 million would probably require 200,000 to 250,000 troops.¹²⁶

However, the idea of using twenty to twenty-five peacekeepers per 1,000 civilians is a historically based rule of thumb, not a binding requirement. To be sure, it is dangerous to ignore rules of thumb in favor of hunches or personal preferences, as has sometimes been done in force planning for specific missions. That said, in a less combustible environment, fewer peacekeepers may suffice. An approach accepting more risk might include stabilization forces in the general range of 100,000, with the U.S. contribution perhaps 30,000 to 50,000.¹²⁷

Consider next the possibility of severe unrest in one of the world’s large countries such as Indonesia, Congo, or Nigeria. At present, such problems are generally seen as of secondary strategic importance to the United States, meaning that Washington may support and help fund a peacekeeping mission under some circumstances but will rarely commit troops—and certainly will not deploy a muscular forcible intervention capability.

However, under some circumstances this situation could change. For example, if al-Qaeda developed a major stronghold in a given large country, the United States might—depending on the circumstances—consider overthrowing the country’s government (if it was in cahoots with terrorists) or helping the government reclaim control over the part of its terroitory occupied by the terrorists. Or it might intervene to help one side in a civil war against another. For example, if the schism between the police and armed forces in Indonesia worsened, and one of the two institutions wound up working with an al-Qaeda offshoot, the United States might accept an invitation from the responsible half of the government to help defeat the other and the terrorist organization in question (this scenario is unlikely today, but perhaps not unthinkable for the future).¹²⁸ Or if a terrorist organization was tolerated in Indonesia, the United States might strike at it directly. That could be the case if the terrorist group took control of land near a major shipping lane in the Indonesian Straits, or simply if it decided to use part of Indonesia for sanctuary.¹²⁹

Clearly, the requirement for foreign forces would be a function of how much of the country in question became unstable, how intact indigenous forces remained, and how large any militia or insurgent force proved to be. For illustrative purposes, if a large fraction of Indonesia or all of Congo were to become ungovernable, the problem could be twice to three times the scale of the Iraq mission. It could be five times the scale of Iraq if it involved trying to restore order throughout Nigeria, though such an operation could be so daunting that a more limited form of intervention seems more plausible—such as trying to stabilize areas where major ethnic or religious groups come into direct contact.

General guidelines for force planning for such scenarios would suggest foreign troop strength up to 100,000 to 200,000 personnel, in rough numbers, based primarily on the populations of the countries in question. That makes them not unlike the scenario of a collapsing or fracturing Pakistan. For these somewhat less urgent missions, compared to those considered in South Asia, U.S. contributions might only be 20–30 percent of the total rather than the 50 percent assumed in the preceding pages. But even so, up to two to three American divisions could be required.

 

QUESTION 7: Could India conquer Pakistan with conventional forces? (A case of Pakistani preemption.)

 

ANSWER: War between these two countries could begin over the disputed region of Kashmir, as it has several times in the past. Or a terrorist group with sanctuary in Pakistan could attack India, as one did several years ago in an assault on parliament, and as occurred again in Mumbai in late 2008. If the attack was serious enough, and was seen as benefiting from even a modest amount of active support from Pakistan’s government, India might consider retaliation with an operation designed to overthrow that government or force its capitulation on terms favorable to India (perhaps involving a complete reversion of Kashmir to India). Knowing this, Pakistan might itself launch a war.

In any event, the military matchup in rough terms is as follows. Pakistan has the equivalent of roughly four ADEs in its military, along with about 100 ground attack aircraft. India has approximately eleven ADEs and about 400 ground attack planes.¹³⁰

In broad terms, India would try to use surprise, advantages in airpower, and maneuver to overwhelm Pakistan. Given India’s substantial numerical advantages, it is very hard to rule out the distinct possibility of a successful Indian attack. Knowing this, perhaps Pakistan would itself preempt India to try to benefit from surprise.

Here is how the mathematics might play out, using the method discussed earlier and based on common elements of the Kugler, Posen, and Epstein approaches. Assume that the combat exchange ratio is even—that is, Pakistan’s advantages of surprise and initiative compensate for India’s advantages of fighting on the defensive. This is a huge assumption and there is no way to be sure it is roughly right in advance. But as a simplifying assumption, it is not a bad place to begin. Assume further that the two air forces perform comparably, managing two sorties a day, delivering on average four munitions per sortie with a 0.1 kill probability per munition, and suffering five percent attrition per sortie due to air defenses (including ground-based systems and enemy air superiority fighters). The casualty math then looks like this, for day one of the war (further assuming a baseline of two percent attrition per day due to ground combat, and an average of 4,000 vehicles per ADE):

 

C (Pakistan)	=	(0.02) × (4 ADEs) = 0.08 ADEs
C (India)	=	C (Pakistan) × (exchange ratio)
	=	C (Pakistan) × 1 = 0.08 ADEs

 

The preceding is from ground combat. Then, adding in the effects of the air war:

 

C (Pakistan)	=	(400 Indian ground-attack planes)
		× (2 sorties/day) × (4 munitions/sortie)
		× (0.1 kill probability/munition)
	=	320 vehicles lost/day = 0.08 ADEs
C (India)	=	(100 Pakistani planes) × (2 sorties/day) × (4) × (0.1)
	=	80 vehicles lost/day = 0.02 ADEs

 

So India’s overall estimated daily losses for day one, theater-wide, would be 0.1 ADEs. Pakistan’s, by contrast, would be 0.16 ADEs. Due to India’s superior air force, Pakistan would lose forces faster than India would, a foreboding development for a country beginning with substantial quantitative inferiority and no hidden trump card (except, alas, its nuclear arsenal).

On subsequent days of battle, the calculations would proceed similarly, though one would first have to adjust for aircraft losses from earlier days of combat. One might also have to adjust for the activation of reserve units and their arrival at the front (though the way I have done this has assumed their mobilization, on both sides, prior to combat). Given the major advantages enjoyed by India, however, details about mobilization would matter little. Unless Pakistan could find a way to benefit from fighting on the offense more than a 1:1 exchange ratio suggests, it would likely find itself quickly in a difficult situation.

 

QUESTION 8: Could an airpower-based defense be used to protect a vulnerable overseas ally?

 

ANSWER: In the modern era, the United States has typically assumed that a country like Kuwait or Saudi Arabia would be difficult to defend from attack (from Saddam’s Iraq, for example, or even from Iran). The basic assumption has been that a given sector of territory might, in effect, have to be initially conceded, then liberated later after a large buildup of some half million U.S. forces dominated by soldiers and Marines. Might it be possible instead to construct a reliable initial defense capability—even without placing huge numbers of American troops on permanent forward deployment to the region in question? If so, this could spare American allies the possible tragedy of even temporary occupation by a hostile foreign power, and could greatly reduce the expense of American military preparations for the scenario in question. This question is not particularly germane for the current Middle East, with tens of thousands of American troops in Iraq and Saddam no longer there, but it could be relevant there or elsewhere in the future.

For a modern military with good intelligence and surveillance capabilities, able to see an enemy coming, defense is—at least in theory—a more favorable form of warfare since it allows one to shoot from protected positions against exposed enemy forces. Of course, being successful on the defense is not a given historically (as noted earlier), so one must carefully assess if intelligence, surveillance, command, and control are up to the job of reliably detecting enemy movements. In addition, the defensive posture works only if a given threat is predictable enough, and a given interest important enough, to warrant deploying substantial amounts of military capability (even if not huge ground forces) in a given theater in advance of a possible crisis.

The airpower component of the Kugler, Posen, and Epstein models can be used to gain at least a very broad sense of the possibilities. Consider a situation like the defense of Saudi Arabia against Iraq. In the 1990s, several RAND analysts built on the success of airpower—and more specifically, precision munitions—in Operation Desert Storm to argue for a different approach to defend Saudi Arabia against attack. Perhaps half as much ground power as used in Desert Storm (or less) might be adequate, provided airpower was fully enabled and exploited. That would have meant purchasing and prestocking large numbers of precision munitions in the theater, ensuring that airplanes had enough refueling aircraft to deploy quickly to the theater as well as bases there from which to operate, and developing optimal reconnaissance capabilities so airpower could be directed to attack moving enemy vehicles before those vehicles could reach their originally intended destinations.¹³¹

The whole scheme may have assumed too much political decisiveness in Washington, Riyadh, and elsewhere about responding to concentrations of Iraqi armored power. It might also have been mistaken to imply that, with a successful defense of Kuwait or Saudi Arabia under its belt, the United States would be content to leave Saddam in power (if it wanted to overthrow him, a larger ground force would presumably have been required regardless). And the concept depended as well on prompt, unfettered access to bases that would have to be available despite the possibility of indigenous country political indecision or debilitating military attack from Saddam’s forces. But the analytical framework was still quite useful for exploring options.

The RAND scholars showed what might be possible using a new type of weapon—specifically, a homing submunition that could find and attack targets largely autonomously. These submunitions, known as SKEETs, can be fired in large numbers. More than 100 can easily be carried on a single tactical combat aircraft (being launched several dozen at a time). If an enemy armored force is sufficiently concentrated, to the point where a single aircraft sortie can achieve multiple expected armor kills, then the mathematics are highly favorable to the defender—provided those munitions, and the planes that carry them, are available in adequate numbers from the opening bell of battle. Nearly 100 Iraqi armored vehicles per day were destroyed during the American air campaign in Desert Storm. It might be possible to be ten times as effective with SKEETs, even if the number of vehicles within the search zone of a given SKEET “swarm” is modest and if multiple SKEETs typically wind up striking the same target (as reflected in the denominator of the expression that follows).¹³²

{[500 ground attack planes][2 sorties/day][100 SKEET/sortie] [0.5 chance of finding targets/sortie] [0.25 probability of kill per SKEET]}/{[10 redundant hits/average destroyed vehicle]} = 1,250 Iraqi vehicles destroyed/day.

Of course, it is very difficult to figure out proper estimates for all the terms in the preceding simple formula. Kill probabilities of munitions can be estimated from performance on the test ranges—but data on such performances is often classified, and in any event results may be much different in a battlefield environment due to enemy countermeasures or other changed conditions. Perhaps even more crucial, and difficult to forecast accurately, is the matter of estimating the probability of detection of targets (as well as the number likely to be within range of a given aircraft all at once). Establishing a basis for determining this figure requires data from similar wars, or detailed technical information about the performance of reconnaissance assets, such as Joint STARS radar-imaging aircraft and similarly equipped unmanned aerial vehicles (UAVs). The ability to estimate such figures reliably depends in large measure on the degree to which analogous wars have been fought in recent times—or at least on the degree to which very realistic simulations and tests have been conducted. So the answer to this question is, perhaps.

 

QUESTION 9: Why do evenly matched militaries sometimes fight to a draw—and sometimes not?

 

ANSWER: The international relations literature is rife with references to balances of power. The casual suggestion of many is that countries or alliance systems that pit roughly comparable military capabilities against each other are less likely to fight, since all would know that any war would likely be long and perhaps inconclusive.

In reality, this is not true. Often, even when two sides are roughly evenly matched according to material indicators, one side wins quickly and decisively. A case in point is Germany’s defeat of France in 1940, but other examples abound.

Of course, the answer has to do with military innovation and entrepreneurship, with surprise and daring, with the precepts of the Chinese tactician Sun Tze rather than the attrition-war mentality of World War I or even the American Civil War. In short, if an attacker such as Germany develops and executes a brilliant plan that its adversary had not contemplated or prepared against, it might win quickly. (An attacker might also overestimate its capabilities and suffer a terrible defeat by striking first, of course.) But how do these general, qualitative observations manifest themselves in the mathematics of the combat equations discussed earlier?

We can use either the Dupuy approach or the Posen/Kugler/Epstein concepts. Assume simply that each side has ten armored divisional equivalents or ADEs (each with an adjusted average of 25,000 troops per ADE), and 500 attack aircraft with an average payload of four munitions flying two sorties a day. This is an oversimplification, but acceptable for my illustrative purposes.

For the Dupuy model, we then need to account for the defensive advantage of the attacked country as well as the surprise advantages of the attacker. Imagine a situation like that of Germany and France in 1940, qualitatively speaking at least. Dupuy’s coefficients of combat limit our choices, but we will assume what are roughly the minimum and maximum here: no advantage for the defender from its preparations, and a 2.5:1 power advantage for the attacker from surprise.¹³³ On top of that, either because of their superior training in general, or because of their meticulous preparation for this particular scenario, the attacking country’s soldiers can be estimated to have twice the quality of the defenders at least for this scenario. All of this would yield, through the power equations:

 

P (attacker)	=	(250,000) × (2.5) × (2) = 1,250,000
P (defender)	=	250,000

 

Daily casualty rates might then be, notionally:

 

C (defender)	=	(0.01) × (250,000) × (5) = 12,500 per day
C (attacker)	=	(0.01) × (250,000) × (1/5) = 500 per day

 

The terms 5 and 1/5 reflect the relative power of the two sides. The defender loss rates are in fact consistent with those of an army that collapses within a couple weeks of hard fighting, given its difficulty of moving reinforcements to (or within) the theater of combat fast enough, as well as the fact that militaries tend to collapse once they have suffered about 50 percent attrition. (To use the other models, we might assume an advantage to the attacker in the combat exchange ratio of as much as ten to one.)

 

QUESTION 10: How could the Persian Gulf oil economy be protected in the face of Iranian efforts to disrupt it?

 

ANSWER: This question resembles the complex analytical challenge of the earlier treatment of a Chinese attack on Taiwan more than the simple mathematical treatments of air-ground combat by Dupuy, Posen, Kugler, and Epstein.

In the 1980s, during the Iran–Iraq War, the United States had to address threats to shipping in the Persian Gulf. To do so, it reflagged some oil tankers under its own colors and enhanced its naval presence in the region.

This type of scenario could recur. But next time, it could do so in a more worrisome way. Given the ongoing state of serious tension in U.S.–Iranian relations, any spark could inflame a serious problem. In recent years, while Iran’s arms imports have not increased as fast as some had feared, they have nonetheless permitted that country to improve its capacity to threaten shipping lanes in the narrow waters of the Persian Gulf and the Strait of Hormuz. In particular, Iran has been improving its capabilities in those very areas of military capability that could cause the United States greatest concern—advanced mines, quiet diesel submarines, and precision-guided antiship missiles.¹³⁴

This hypothetical worry could become acute, for example, if in the coming years Israel or the United States attacked the exposed parts of Iran’s nuclear infrastructure. In such an event, the United States might reinforce its defensive position in the region in advance. But an aggressive Iranian response against American friends and allies in the region, or against oil tankers in the Persian Gulf, could result anyway.

To defend the Gulf, reconnaissance and rapid-strike capabilities could be provided either via sea-based assets or land-based capabilities. Aerial and sea reconnaissance, as well as quick-strike capabilities, would be needed. American submarines would probably be desired to keep a constant track on Iranian submarines. And of course, ships to protect convoys would likely be required as well, as could mine warfare vessels.

The quantitative requirements for these various assets would be a function of three main sets of factors: geography, rotational policies, and total Iranian force strength. The United States, and any assisting allies, would need to maintain robust quick-action capabilities along the whole length of the Gulf. It would need to be able to sustain coverage twenty-four hours a day, and it would need to be able to face down an all-out Iranian assault if necessary as well.

In rough terms, these sizing criteria lead to the following rough requirements. Given Iran’s small submarine force, with just three vessels, the demand for American forces would probably not require more than twenty submarines (allowing up to two U.S. subs per Iranian submarine as well as the need to rotate American ships). Even this is a very conservative estimate, since it is not apparent how functional these Iranian submarines truly are, and intelligence reports cast doubt on their likely effectiveness.

To ensure continual airspace dominance in the Gulf, roughly as many planes could be required as were needed to enforce the northern and southern no-fly zones over Iraq from 1991 to 2003—some 200 planes in all. (The size of the Persian Gulf is roughly comparable to that of Iraq, if one includes littoral regions as well as the body of water itself, so this is a reasonable approximation.) The aircraft would ideally be based at several locations along the 500 mile length of the Gulf to minimize time wasted in transit and allow for rapid reinforcement should Iran attempt an assault. In addition, some additional number of planes might need to be capable of rapid response in the face of any Iranian aerial action. (Iran’s total air force numbers about 300 planes, of which perhaps 200 are airworthy.)¹³⁵

Enough surveillance aircraft would be needed to maintain orbits at the northern and southern ends of the Gulf. Allowing for rest time for crews and maintenance time for planes, that makes for a grand total of eight to ten planes for air monitoring and a similar number for sea surveillance. (The need for two separate “orbits” of reconnaissance aircraft again accords with the Iraq no-fly-zone experience, as well as the fact that radar horizons of aircraft at 35,000 to 40,000 feet are typically about 250 miles.)¹³⁶

In addition to its convoy escorts the United States might need to create a “fence” of ships capable of ballistic-missile defense spaced every fifty miles along Iran’s seacoast to ensure short enough reaction times to any missile launch. The spacing follows from the fact that Iran’s ballistic missiles would only have to travel a comparable distance, suggesting a very short flight time. Defensive interceptors would not travel notably faster than the incoming threats. There would also be a delay between the launch of the ballistic missiles and the launch of interceptor as threats were recognized and command decisions were made.

Taken together, the preceding assets resemble the air and naval components of what has commonly been considered a one-war force package in recent times. Whether some ground forces were needed as a prudent deterrent against overland Iranian aggression would also have to be considered, but the numbers here would presumably not have to reach into the “major theater war” magnitudes.

NOTES

1. Geoffrey Blainey, The Causes of War (New York: The Free Press, 1973), pp. 246–49.

2. By “roughly comparable military capabilities,” I mean a situation in which neither side is estimated to exceed the other’s capabilities by more than 50 percent. See Requirements and Resources Directorate, U.S. Army Concepts Analysis Agency, Combat History Analysis Study Effort (CHASE): Progress Report (Washington, D.C.: Army Concepts Analysis Agency, 1986), p. 3–20, cited in Joshua M. Epstein, “Dynamic Analysis and the Conventional Balance in Europe,” International Security, vol. 12, no. 4 (Spring 1988), p. 156.

3. Richard K. Betts, Surprise Attack (Washington, D.C.: Brookings, 1982), pp. 5–16.

4. Epstein, “Dynamic Analysis and the Conventional Balance in Europe,” p. 156.

5. Eric V. Larson, Casualties and Consensus (Santa Monica, Calif.: RAND, 1996).

6. See Joshua M. Epstein, Measuring Military Power: The Soviet Air Threat to Europe (Princeton, N.J.: Princeton University Press, 1984), pp. xxv–xxvi.

7. For a good explanation of TACWAR, see Francis P. Hoeber, Military Applications of Modeling: Selected Case Studies (New York: Gordon and Breach Science Publishers, 1981), pp. 132–42.

8. Michael R. Gordon and General Bernard E. Trainor, The Generals’ War: The Inside Story of the Conflict in the Gulf (Boston, Mass.: Little, Brown, and Co., 1995), p. 457; Department of Defense, Conduct of the Persian Gulf War: Final Report to Congress (Washington, D.C.: Department of Defense, April 1992), pp. A-3 through A-11; and Congressional Budget Office, “Costs of Operation Desert Shield,” CBO Staff Memorandum,” U.S. Congress, Washington, D.C., January 1991, p. 15.

9. See, for example, John J. Mearsheimer, “Why the Soviets Can’t Win Quickly in Central Europe,” International Security, vol. 7, no. 1 (Summer 1982), reprinted in Steven E. Miller, ed., Conventional Forces and American Defense Policy (Princeton, N.J.: Princeton University Press, 1986), pp. 121–57; Barry R. Posen, “Measuring the European Conventional Balance: Coping with Complexity in Threat Assessment,” International Security, vol. 9, no. 3 (Winter 1984/85), reprinted in Miller, ed., Conventional Forces and American Defense Policy, pp. 79–120; Joshua M. Epstein, “Dynamic Analysis and the Conventional Balance in Europe,” International Security, vol. 12, no. 4 (Spring 1988), pp. 154–65; Eliot A. Cohen, “Toward Better Net Assessment: Rethinking the European Conventional Balance,” International Security, vol. 13, no. 1 (Summer 1988), pp. 50–89; Steven J. Zaloga and Malcolm Chalmers, “Is There a Tank Gap?: Comparing NATO and Warsaw Pact Tank Fleets,” International Security, vol. 13, no. 1 (Summer 1988), pp. 5–49; and Lutz Unterseher, “Correspondence: The Tank Gap Data Flap,” International Security, vol. 13, no. 4 (Spring 1989), pp. 180–87.

10. FEBA stands for forward edge of the battle area.

11. For a good explanation and critique of the Lanchester equations, see Joshua M. Epstein, Strategy and Force Planning: The Case of the Persian Gulf (Washington, D.C.: Brookings Institution, 1987), pp. 146–55.

12. For an explanation of the advantages of simpler, more transparent models, see Zalmay Khalilzad and David Ochmanek, “Rethinking US Defence Planning,” Survival, vol. 39, no. 1 (Spring 1997), pp. 43–64.

13. See Barry R. Posen, “Measuring the European Conventional Balance: Coping with Complexity in Threat Assessment,” International Security, vol. 9, no. 3 (Winter 1984/85), reprinted in Steven E. Miller, ed., Conventional Forces and American Defense Policy (Princeton, N.J.: Princeton University Press, 1986), pp. 79–120.

14. Posen, “Measuring the European Conventional Balance,” p. 106; and William P. Mako, U.S. Ground Forces and the Defense of Central Europe (Washington, D.C.: Brookings, 1983), pp. 36–37.

15. To compute ADE scores for various militaries in order to make these ADE comparisons, a system known as the WEI-WUV method is often employed. American qualitative advantages show up partly in the ADE scores, but even more in the exchange ratio. See U.S. Army Concepts Analysis Agency, War Gaming Directorate, Weapon Effectiveness Indices/Weighted Unit Values III (Bethesda, Md.: CAA, 1979).

It is also possible, however, to use a modified approach to scoring static inputs. Rather than employ the WEI-WUV system, a method such as TASCFORM can be employed (the name of which derives from The Analytical Sciences Corporation, which created this database or formula). TASCFORM shows a much greater advantage for Western equipment over alternatives such as Soviet weaponry. In such a situation, if TASCFORM is used rather than WEI-WUV scoring, the exchange ratio would be less lopsided in the U.S. favor (since much of the American advantage would have already been captured in the weapons input scores).

See Lane Pierrot, Structuring U.S. Forces After the Cold War: Costs and Effects of Increased Reliance on the Reserves (Washington, D.C.: Congressional Budget Office, 1992), pp. 46–53; and Michael E. O’Hanlon, The Art of War in the Age of Peace: U.S. Military Posture for the Post–Cold-War World (Westport, Conn.: Praeger, 1992), p. 67.

16. Joshua M. Epstein, “Dynamic Analysis and the Conventional Balance in Europe,” International Security, vol. 12, no. 4 (Spring 1988), pp. 155–58; and Joshua M. Epstein, Conventional Force Reductions: A Dynamic Assessment (Washington, D.C.: Brookings, 1990), pp. 51–65.

17. See, for example, Epstein, Strategy and Force Planning, pp. 63–88, 117–25; and Joshua M. Epstein, Conventional Force Reductions: A Dynamic Assessment (Washington, D.C.: Brookings, 1990), pp. 48–80.

18. It is worth noting that once breakthroughs occur, motorized armies can generally advance ten to sixty kilometers a day depending on factors like terrain and any residual resistance (in fact, even in the nineteenth century, armies sometimes averaged ten to twenty kilometers of progress a day). Against strong resistance, attackers more frequently average moving one to five kilometers a day depending on the quality and preparedness of the defense and related factors. See Trevor N. Dupuy, Numbers, Predictions, and War: The Use of History to Evaluate and Predict the Outcome of Armed Conflict, revised edition (Fairfax, Va.: HERO Books, 1985), pp. 16, 213–14; and Jeffrey Record, “Armored Advance Rates: A Historical Inquiry,” Military Review, vol. 53, no. 9 (September 1973), pp. 63–66.

19. See Tami Davis Biddle, Rhetoric and Reality in Air Warfare: The Evolution of British and American Ideas about Strategic Bombing, 1914–1945 (Princeton, N.J.: Princeton University Press, 2002), pp. 289–301.

20. Thomas A. Keaney and Eliot A. Cohen, Gulf War Air Power Survey Summary Report (Washington, D.C.: Government Printing Office, 1993), pp. 65, 199; and General Accounting Office (now the Government Accountability Office), Operation Desert Storm: Evaluation of the Air Campaign (Washington, D.C.: GAO, June 1997), GAO/NSIAD-97-134, p. 166.

21. Benjamin S. Lambeth, NATO’s Air War for Kosovo: A Strategic and Operational Assessment (Santa Monica, Calif.: RAND, 2001), pp. 35, 62, 65; and Ivo H. Daalder and Michael E. O’Hanlon, Winning Ugly: NATO’s War to Save Kosovo (Washington, D.C.: Brookings, 2000), pp. 135–36.

22. Trevor N. Dupuy, Attrition: Forecasting Battle Casualties and Equipment Losses in Modern War (Fairfax, Va.: HERO Books, 1990), p. 139.

23. Barry R. Posen, “Political Objectives and Military Options in the Persian Gulf,” Defense and Arms Control Studies Working Paper, Massachusetts Institute of Technology, Cambridge, Mass. (November 1990), pp. 24–25.

24. Joshua M. Epstein, “War with Iraq: What Price Victory?” Briefing Paper, Brookings Institution, December 1990.

25. Michael R. Gordon and Bernard E. Trainor, The Generals’ War: The Inside Story of the Conflict in the Gulf (Boston, Mass.: Little, Brown, and Co., 1995), pp. 355–80.

26. Congressional Budget Office, “Costs of Operation Desert Shield,” January 1991, p. 15.

27. Dupuy, Attrition, pp. 73–74, 131.

28. See Directorate for Information Operations and Reports, “Persian Gulf War: Desert Shield and Desert Storm,” Department of Defense, Dec. 15, 2001 (webi.whs.osd.mil/mmid/casualty); DoDefense, Conduct of the Persian Gulf War: Final Report to Congress (April 1992), p. M-I.

29. See also, Lawrence Freedman and Efraim Karsh, The Gulf Conflict, 1990–1991: Diplomacy and War in the New World Order (Princeton, N.J.: Princeton University Press, 1993), p. 409.

30. Stephen Biddle, “Victory Misunderstood: What the Gulf War Tells Us about the Future of Conflict,” International Security, vol. 21, no. 2 (Fall 1996), pp. 139–79

31. JSTARS was first used in Desert Storm; it can scan a region of twenty or more kilometers on a side when in broad-sweep mode. See James F. Dunnigan, How to Make War: A Comprehensive Guide to Modern Warfare for the Post–Cold War Era (New York: William Morrow and Company, Inc., 1993), pp. 154–55.

32. Stephen Biddle, “The Past as Prologue: Assessing Theories of Future Warfare,” Security Studies, vol. 8, no. 1 (Autumn 1998), pp. 1–

33. For the sake of reference, Iraq’s army of the time of roughly one million soldiers had about fifty divisions, whereas the U.S. Army of roughly 1.8 million total soldiers had only two-thirds as many, counting National Guard formations. The United States organizes a larger share of its soldiers into nondivisional support units than do many other militaries. See International Institute for Strategic Studies, The Military Balance 1989–1990 (London: Brassey’s, 1989), pp. 16–18, 101.

For reference purposes, while U.S. divisions typically have 16,000 to 18,000 soldiers, brigades have 2,000 to 5,000, battalions 250 to 1,000, companies 100 to 300, platoons 30 to 75, and squads 4 to 13. Each unit tends to have about three to five of the smaller echelon unit within it; there are three to five squads in a platoon, three to five platoons in a company, three to five companies in a battalion, five or six battalions in a brigade, and three or four brigades in a division (the terminology changes somewhat for the U.S. Marine Corps, with regiments being the closest approximation to brigades). See Robert M. Perito, ed., Guide for Participants in Peace, Stability, and Relief Operations (Washington, D.C.: U.S. Institute of Peace, 2007), p. 251.

Looked at another way, in the U.S. Army of 2005, there were ten active divisions and eight reserve component divisions (all the latter in the National Guard). There were also thirty-three active brigades and thirty-six reserve component brigades (most within the divisional structures). When the Army reaches its intended size of forty-three active brigades, there will also be thirty-four reserve component brigades, and if the process continues to the maximum extent now being considered, there will be forty-eight active brigades and thirty-four reserve component brigades. (Put in other terms, there were 98 active battalions and 108 reserve component battalions in 2005. The plan was to wind up with 92 active battalions and 70 reserve battalions initially, with the possibility of then going to 102 active battalions and 70 reserve battalions.) In terms of companies, the 2005 figures were 297 and 327; the interim targets are 353 and 265; the 48-brigade force plans are for 393 and 265, respectively. See Adam Talaber, Options for Restructuring the Army (Washington, D.C.: Congressional Budget Office, 2005), pp. 9, 15.

34. See Thomas A. Keaney and Eliot A. Cohen, Gulf War Air Power Survey Summary Report (Washington, D.C.: Government Printing Office, 1993), pp. 21, 58–64, 155; and Les Aspin and William Dickinson, Defense for a New Era: Lessons of the Persian Gulf War (Washington: Brassey’s, 1992), pp. 1–41; for data on Arab–Israeli wars, see Posen, “Measuring the Europe an Conventional Balance,” in Miller, Conventional Forces and American Defense Policy, p. 113; and Dupuy, Numbers, Predictions, and War, pp. 118–39.

35. Other work, such as that done at the RAND Corporation, improves the inputs used in the air-only parts of the models. See Christopher Bowie, Fred Frostic, Kevin Lewis, John Lund, David Ochmanek, and Philip Propper, The New Calculus (Santa Monica, Calif.: RAND, 1993); David A. Ochmanek, Edward R. Harshberger, David E. Thaler, and Glenn A. Kent, To Find, and Not to Yield (Santa Monica, Calif.: RAND, 1998).

36. As noted before, it is also possible to use a modified approach to scoring static inputs. Rather than employ the WEI-WUV system, a method such as TASCFORM can be employed (whose name, again, derives from The Analytical Sciences Corporation, which created this database or formula). TASCFORM shows a much greater advantage for Western equipment over alternatives such as Soviet weaponry. It can further be used to reflect the varying degrees of skill among soldiers on various ides through its personnel weighting system. This approach is useful when trying to create a static indicator that more accurately gauges overall combat power. However, in a dynamic model, it is unnecessary, and confusing in some ways, since the issue is less about who has the better equipment and much more about who is better at employing it. (Those who doubt this might be reminded that the so-called “opposition forces” operating at America’s combat training centers–American soldiers using Russian weaponry–typically defeat the visiting main U.S. combat forces in simulated battle. In addition, as Stephen Biddle has shown for the case of Operation Desert Storm, Iraqi defeat had far more to do with poor tactics and poor use of equipment than with the quality of the equipment per se.)

See Lane Pierrot, Structuring U.S. Forces After the Cold War: Costs and Effects of Increased Reliance on the Reserves (Washington, D.C.: Congressional Budget Office, 1992), pp. 46–53; and Michael E. O’Hanlon, The Art of War in the Age of Peace: U.S. Military Posture for the Post–Cold-War World (Westport, Conn.: Praeger, 1992), p. 67.

37. Barry R. Posen, “Measuring the Europe an Conventional Balance: Coping with Complexity in Threat Assessment,” International Security, vol. 9, no. 3 (Winter 1984/1985), reprinted in Steven E. Miller, ed., Conventional Forces and American Defense Policy (Princeton, N.J.: Princeton University Press, 1986), p. 105.

38. Keaney and Cohen, Gulf War Air Power Survey Summary Report, pp. 105–6; and General Accounting Office, Operation Desert Storm: Evaluation of the Air Campaign, GAO/NSIAD-97-134, pp. 8–10, 105–7, 146–48, 157–59.

39 Civilian casualty estimates based on the briefing by William Arkin of Greenpeace to Gulf War Air Power Survey project members, Oct. 31, 1991, cited in Keaney and Cohen, Gulf War Air Power Survey Summary Report, p. 75; for military casualty estimates, see Keaney and Cohen, p. 107.

40. Keaney and Cohen, Gulf War Air Power Survey Summary Report, pp. 104–17, 203.

41. Posen, “Measuring the Europe an Conventional Balance,” p. 104.

42. Joshua M. Epstein, Strategy and Force Planning: The Case of the Persian Gulf (Washington, D.C.: Brookings, 1987), p. 113; and Frances M. Lussier, Replacing and Repairing Equipment Used in Iraq and Afghanistan: The Army’s Reset Program (Washington, D.C.: Congressional Budget Office, 2007), p. xi.

43. Lane Pierrot, Planning for Defense: Affordability and Capability of the Administration’s Program (Washington, D.C.: Congressional Budget Office, 1994), p. 22; William W. Kaufmann, Assessing the Base Force: How Much Is Too Much? (Washington, D.C.: Brookings, 1992), pp. 52–56; and Steven R. Bowman, “Persian Gulf War: Summary of U.S. and Non-U.S. Forces,” Congressional Research Service, February 11, 1991, pp. 1–8.

44. These scores are based on the TASCFORM system of scoring more than the WEI-WUV system. See Michael E. O’Hanlon, The Art of War in the Age of Peace (Westport, Conn.: Praeger, 1992), p. 67.

45. Posen, “Measuring the Europe an Conventional Balance,” in Miller, ed., Conventional Forces and American Defense Policy, p. 133.

46. Stephen Biddle, Military Power: Explaining Victory and Defeat in Modern Battle (Princeton, N.J.: Princeton University Press, 2004), p. 1.

47. Historically, air-to-air combat causes substantial losses, too. As an example, the United States lost 30 percent of all aerial encounters during the first three years of the Vietnam War. It benefited from advantages of about 10:1 in the second half of the Vietnam conflict, as it had in Korea as well as the Pacific campaign of World War II. Israel achieved aerial exchange ratios of at least 20:1 in its 1967, 1973, and 1982 engagements with Arab states. See Joshua M. Epstein, Measuring Military Power: The Soviet Air Threat to Europe (Princeton, N.J.: Princeton University Press, 1984), pp. 110–12. On Desert Storm, see Thomas A. Keaney and Eliot A. Cohen, Gulf War Air Power Survey Summary Report (Washington, D.C.: Government Printing Office, 1993), p. 61.

48. Keaney and Cohen, Gulf War Air Power Survey, p. 211.

49. Barry D. Watts, The Military Uses of Space: A Diagnostic Assessment (Washington, D.C.: Center for Strategic and Budgetary Assessments, 2001), pp. 42–3; Anne Marie Squeo, “U.S. Military’s GPS Reliance Makes a Cheap, Easy Target,” The Wall Street Journal, Sept. 24, 2002.

50. Keaney and Cohen, Gulf War Air Power Survey Summary Report, p. 173.

51. Biddle, “Victory Misunderstood,” pp. 166–68.

52. See Trevor N. Dupuy, Attrition: Forecasting Battle Casualties and Equipment Losses in Modern War (Fairfax, Va.: HERO Books, 1990), pp. 104–32; see also Trevor N. Dupuy, Numbers, Predictions, and War, revised edition (Fairfax, Va.: HERO Books, 1985); and Trevor N. Dupuy, Understanding War: History and Theory of Combat (New York: Paragon Books, 1987).

53. Dupuy, Attrition, p. 76.

54. Ibid., Attrition: Forecasting Battle Casualties and Equipment Losses in Modern War (Fairfax, Va.: HERO Books, 1990), pp. 146–51.

55. ———, If War Comes, How to Defeat Saddam Hussein (Fairfax, Va.: HERO Books, 1991), p. 104; Congressional Budget Office, “Costs of Operation Desert Shield,” Jan. 1991, p. 15.

56. See Robert L. Goldich, “Casualties and Maximum Number of Troops Deployed in Recent U.S. Military Ground Combat Actions,” Congressional Research Service, Oct. 8, 1993; Brig. Gen. Robert H. Scales, Certain Victory: The U.S. Army in the Gulf War (Washington, D.C.: Brassey’s, 1994), pp. 32–35; International Institute for Strategic Studies, The Military Balance 1989–1990 (Washington, D.C.: Brassey’s, 1989), pp. 26–27, 199; and Susan L. Marquis, Unconventional Warfare: Rebuilding U.S. Special Operations Forces (Washington, D.C.: Brookings, 1997), pp. 187–201.

57. See Statement of General James R. Harding, Director, Inter-American Region, Office of the Secretary of Defense, before the Subcommittee on Western Hemi sphere Affairs of the House Foreign Affairs Committee, July 30, 1991 (www.nexis.com/research/search/submitViewTagged).

58. I have simplified Dupuy’s method considerably. He employs factors to account for terrain, surprise, weather, and so on. More important, the way in which relative power differentials enter into his equations is not quite linear in the way I have suggested. But his method has an arbitrary quality about it at times as well–for example, he adds a “sophistication factor” in addition to mobility, firepower, and combat effectiveness coefficients to account for the quality of one military over another. It is unclear why so many different such factors are needed to explain similar phenomena, or how one selects the proper value for each. By contrast, his methodology for computing power is relatively straightforward. For more exact information on how power ratios enter into his calculations, see Dupuy, Attrition, pp. 124–27, 146–52.

59. That is, 250,000 times their quality advantage of about three times a benefit from surprise of about 10 percent.

60. That is, 100,000 troops times a factor of two advantage due to the benefits of being on the defensive and fighting within a city.

61. The relative power term would not vary quite this much from one country to the other according to Dupuy’s detailed tables, but for simplicity of use and for gaining an approximate sense of the calculations, these figures are not far off.

62. Dupuy, Attrition: Forecasting Battle Casualties and Equipment Losses in Modern War (Fairfax, Va.: HERO Books, 1990), p. 149.

63. Michael O’Hanlon, “Estimating Casualties in a War to Overthrow Saddam,” Orbis, vol. 47, no. 1 (Winter 2003), pp. 36–37.

64. Some of my ideas here first appeared in Michael O’Hanlon, “Why China Cannot Conquer Taiwan,” International Security, vol. 25, no. 2 (Fall 2000), pp. 51–86.

65. See U.S. Marine Corps, “Operational Maneuver from the Sea,” Marine Corps Gazette, June 1996.

66. As noted airpower expert Richard Hallion writes, “Air superiority characterizes a war where a nation can exert its power over a foe with few air losses of its own, and without serious concern about the enemy’s ability to contest for control of the air with its own air forces. The foe, suffering from air inferiority, can only undertake limited offensive action, and must devote the bulk of activity to defensive warfare. . . . Air supremacy implies that a nation can control a foe with essentially no or absolutely minimal air losses of its own, and without need to concern itself about the enemy’s air intentions.” See Richard P. Hallion, “Control of the Air: The Enduring Requirement,” Air Force History and Museums Program, Bolling Air Force Base, Washington, D.C., September 1999, available at www.af.mil/shared/media/document/AFD-060726-027.pdf [accessed March 14, 2008].

67. Office of the Secretary of Defense, “Annual Report to Congress: Military Power of the People’s Republic of China 2008,” Department of Defense, Washington, D.C., 2008, p. 2.

68. Ministry of National Defense, Republic of China (Taiwan), 2004 National Defense Report (Taipei, Taiwan: Ministry of National Defense, 2004), p. 29.

69. Epstein, Measuring Military Power, pp. 208–9, 223.

70. Personal communication from Shuhfan Ding, Institute of International Relations, National Chengchi University, Taipei, Taiwan, April 14, 2000; see also David Shambaugh, “China’s Military Views the World,” International Security, vol. 24, no. 3 (Winter 1999/2000), p. 61.

71. See, for example, David Shlapak, “Projecting Power in a China-Taiwan Contingency: Implications for USAF and USN Collaboration,” in Stuart E. Johnson and Duncan Long, eds., Coping with the Dragon: Essays on PLA Transformation and the U.S. Military (Washington, D.C.: National Defense University, December 2007), p. 90.

The U.S. experience against Iraq in Desert Storm in 1991 provides a good window into how hard it is to shut down an enemy’s air force. Coalition aircraft averaged dozens of strike sorties a day against Iraqi airfields during the war’s first week, yet this did not stop the Iraqi air force from flying about forty sorties a day. That was at a time when coalition aircraft completely ruled the skies, moreover. In the airfield attacks, British planes were dropping advanced runway-penetrating weapons, precisely and from low altitude. They carried some thirty bomblets apiece, each bomblet consisting of two charges: a primary explosive to create a small hole in the runway, and a second explosive to detonate below its surface, causing a crater often to twenty meters’ width (depending largely on soil conditions). A standard attack would have used eight aircraft, each dropping two weapons, to shut down a standard NATO-length runway of 9,000 feet by 150 feet–a difficult mission, given the need to drop the weapons at precise and quite low altitudes.

See Keaney and Cohen, Gulf War Air Power Survey Summary Report (Washington, D.C.: Government Printing Office, 1993), pp. 56–65; General Accounting Office, Operation Desert Storm: Evaluation of the Air Campaign GAO/NSIAD-97-134 (June 1997), pp. 209–12; and Christopher S. Parker, “New Weapons for Old Problems,” International Security, vol. 23, no. 4 (Spring 1999), p. 147; Duncan Lennox, ed., Jane’s Air-Launched Weapons (Surrey, England: Jane’s Information Group, 1999), issue 33 (August 1999); and Christopher M. Centner, “Ignorance Is Risk: The Big Lesson from Desert Storm Air Base Attacks,” Airpower Journal, vol. VI, no. 4, (Winter 1992), pp. 25–35 [available at www.airpower.maxwell.af.mil/airchronicles/apj/centner.html]; and personal communication from Dave C. Fidler, Wing Commander Air 1, British Embassy, Washington, D.C., April 14, 2000.

72. See Department of Defense, FY04 Report to Congress on PRC Military Power: Annual Report on the Military Power of the People’s Republic of China (2004), p. 40; and International Institute for Strategic Studies, The Military Balance 2005/2006 (Colchester: Routledge, 2005), pp. 270–76.

73. See James F. Dunnigan, How to Make War: A Comprehensive Guide to Modern Warfare for the Post–Cold War Era, 3rd ed. (New York: William Morrow and Company, Inc., 1993), pp. 284–92.

74. Wendell Minnick, “Washington Establishes Military Hotline with Taipei,” Jane’s Defence Weekly, October 29, 2003, p. 14.

75. Swaine, Taiwan’s National Security, Defense Policy, and Weapons Procurement Processes, p. 60.

76. Robert Hewson, “China Boosts Its Air Assets with Ilyushin Aircraft,” Jane’s Defence Weekly, September 21, 2005, p. 16.

77. For historical perspective, see James A. Huston, “The Air Invasion of Holland,” Military Review (September 1952), pp. 13–27; and Gerard M. Devlin, Paratrooper!: The Saga of U.S. Army and Marine Parachute and Glider Combat Troops During World War II (New York: St. Martin’s Press, 1979).

78. This is a simplified version of Hughes’ formula; see Hughes, Fleet Tactics and Coastal Combat, p. 268.

79. Capt. Wayne P. Hughes, Jr. (U.S. Navy, retired), Fleet Tactics and Coastal Combat, 2nd edition (Annapolis, Mary land: Naval Institute Press, 2000), pp. 268–79.

80. Nordeen, Air Warfare in the Missile Age, pp. 201–3; and Jane’s Naval Review 1987 (London: Jane’s Publishing, 1987), p. 124. In flying some 300 sorties against British forces in the Falklands War, and attacking the UK ships on the open oceans where they are harder to spot than when approaching shore, Argentina sank four British ships with bombs and hit another six with bombs that did not detonate because they had been improperly fused. Argentina sank a total of six ships including those hit by Exocets and other weapons.

81. International Institute for Strategic Studies, The Military Balance 1981/1982 (London: International Institute for Strategic Studies, 1982), pp. 92–93.

82. Ship losses are discussed in the text. As for aircraft attrition, Taiwan has well over 100 surface-to-air missile batteries with ranges oftens of kilometers–more than enough to have some coverage near all of its twenty to thirty large airfields and five major ports (the kinds of places where PRC paratroopers might do the most good, seizing assets that could then be used to deploy PRC reinforcements). In addition to its air force, it also has hundreds of anti-aircraft guns and many smaller surface-to-air missile batteries that use high-quality modified Sidewinder and Sparrow missiles.

83. Office of Technology Assessment, Proliferation of Weapons of Mass Destruction (Washington, D.C.: Office of Technology Assessment, 1993), pp. 45–67.

84. Dupuy, Attrition: Forecasting Battle Casualties and Equipment Losses in Modern War (Fairfax, Va.: HERO Books, 1990), p. 58; and Anthony H. Cordesman and Abraham R. Wagner, Lessons of Modern War, Volume 2: The Iran–Iraq War (Boulder, Colo.: Westview Press, 1990), p. 518.

85. See Utgoff, The Challenge of Chemical Weapons, pp. 148–88; and Dupuy, Attrition, p. 58.

86. Robert G. Nagler, Ballistic Missile Proliferation: An Emerging Threat (Arlington, Va.: System Planning Corporation, 1992), p. 10.

87. See Dunnigan, How to Make War, pp. 284–92. The typical lateral inaccuracy of gunfire or artillery fire is proportional to the distance over which the round must travel, meaning that a shot to 500 meters would be expected to have one-tenth the miss distance of a shot to five kilometers.

88. For a concurring view, see McVadon, “PRC Exercises, Doctrine, and Tactics Toward Taiwan,” pp. 254–55.

89. Based on the Iran–Iraq War experience, as well as basic ballistics and blast information, warheads from a missile with several hundred kilograms of explosive might kill ten to twenty people on average, in a typical city, though the potential exists for much more devastation depending on where the explosion occurs. Artillery and mortar rounds, with warheads about one-tenth that size (say seven to twenty-five pounds of explosive, as noted here) would have about one-fifth the lethal effect. (The lethal area of an explosive varies roughly with the two-thirds power of the explosive yield. So an eightfold reduction in power produces a fourfold decrease in lethal area–and hence expected casualties.) The lethal radii for weapons in this range could be thirty to fifty meters for mortar rounds and artillery shells, and 100 meters or more for typical missile warheads. See Anthony H. Cordesman and Abraham R. Wagner, The Lessons of Modern War, Volume II: The Iran–Iraq War (London: Westview Press, 1990), pp. 364–68; U.S. Army, Field Manual 5-34: Engineer Field Data (Washington, D.C., September 1987), p. 4-1; James F. Dunnigan, How to Make War: A Comprehensive Guide to Modern Warfare for the Post–Cold War Era (New York: William Morrow and Company, Inc., 1993), pp. 124–25; and Janne E. Nolan, Trappings of Power: Ballistic Missiles in the Third World (Washington, D.C.: Brookings, 1991), pp. 68–69. I also thank Colonel Thomas Lynch of the U.S. Army for information on artillery via private correspondence, March 17, 2008.

90. John Hill, “Missile Race Heightens Tension Across Taiwan Strait,” Jane’s Intelligence Review (January 2005), pp. 44–45.

91. Anthony H. Cordesman and Abraham R. Wagner, The Lessons of Modern War, Volume II: The Iran–Iraq War (Boulder, Colo.: Westview Press, 1990), pp. 205–6; and Daniel L. Byman and Matthew C. Waxman, “Kosovo and the Great Air Power Debate,” International Security, vol. 24, no. 4 (Spring 2000), pp. 37–38.

92. James C. Mulvenon, Murray Scot Tanner, Michael S. Chase, David Frelinger, David C. Gompert, Martin C. Libicki, and Kevin L. Pollpeter, Chinese Responses to U.S. Military Transformation and Implications for the Department of Defense (Santa Monica, Calif.: RAND Corporation, 2006), pp. 116–20.

93. John Hill, “Missile Race Heightens Tension Across Taiwan Strait,” Jane’s Intelligence Review (January 2005), pp. 44–45.

94. Bernard D. Cole, “Right-Sizing the Navy: How Much Naval Force Will Beijing Deploy?” in Roy Kamphausen and Andrew Scobell, eds., Right-Sizing the People’s Liberation Army: Exploring the Contours of China’s Military (Carlisle, Pa.: Strategic Studies Institute, Army War College, 2007), pp. 541–42.

95. For discussion of Chinese writings that seem to take a similar tack, see Roger Cliff, Mark Burles, Michael S. Chase, Derek Eaton, and Kevin L. Pollpeter, Entering the Dragon’s Lair: Chinese Antiaccess Strategies and Their Implications for the United States (Santa Monica, Calif.: RAND, 2007), pp. 66–73. Among other naval force modernizations, China now has about twenty modern attack submarines in its fleet, and it is also expected to acquire ocean reconnaissance satellites (early versions of which it already reportedly possesses) as well as communications systems capable of reaching deployed forces in the field in the next five to ten years. See Office of the Secretary of Defense, Military Power of the People’s Republic of China, 2008: Annual Report to Congress (Washington, D.C.: Department of Defense, 2008), available at www.defenselink.mil/pubs/pdfs/China_Military_Report_08.pdf, pp. 4, 27 [accessed March 20, 2008]; and Michael McDevitt, “The Strategic and Operational Context Driving PLA Navy Building,” in Roy Kamphausen and Andrew Scobell, eds., Right-Sizing the People’s Liberation Army: Exploring the Contours of China’s Military (Carlisle, Pa.: Strategic Studies Institute, Army War College, 2007), p. 499.

96. Capt. Wayne P. Hughes, Jr. (U.S. Navy, retired), Fleet Tactics and Coastal Combat, 2nd edition (Annapolis, Mary land: Naval Institute Press, 2000), pp. 224–27.

97. O’Hanlon, Defense Policy Choices for the Bush Administration, p. 189.

98. See Congressional Budget Office, U.S. Naval Forces: The Sea Control Mission (Washington, D.C.: Congressional Budget Office, 1978).

99. For analyses that remain mostly correct today, see Owen Cote, The Future of Naval Aviation (Cambridge, Mass.: MIT Security Studies Program, 2006), pp. 34–37, available at http://web.mit.edu/ssp/; Owen Cote and Harvey Sapolsky, Antisubmarine Warfare After the Cold War (Cambridge, Mass.: MIT Security Studies Program, 1997), p. 13; and Tom Stefanick, Strategic Antisubmarine Warfare and Naval Strategy (Lexington, Mass.: Lexington Books, 1987), pp. 35–49.

100. Capt. Wayne P. Hughes, Jr. (U.S. Navy, retired), Fleet Tactics and Coastal Combat, 2nd edition (Annapolis, Mary land: Naval Institute Press, 2000), pp. 172–73.

101. Desmond Ball, “China Pursues Space-Based Intelligence Gathering Capabilities,” Jane’s Intelligence Review (December 2003), pp. 36–39.

102. Michael E. O’Hanlon, Neither Star Wars Nor Sanctuary: Constraining the Military Uses of Space (Brookings, 2004), pp. 91–104; Bill Gertz, “Chinese Missile Has Twice the Range U.S. Anticipated,” The Washington Times, November 20, 2002, p. 3; Barry Watts, The Military Uses of Space: A Diagnostic Assessment (Washington, D.C.: Center for Strategic and Budgetary Assessments, 2001); Bob Preston, Dana J. Johnson, Sean J. A. Edwards, Michael Miller, and Calvin Shipbaugh, Space Weapons, Earth Wars (Santa Monica, Calif.: RAND, 2002); and Benjamin Lambeth, Mastering the Ultimate High Ground: Next Steps in the Military Uses of Space (Santa Monica, Calif.: RAND, 2003).

103. For one good concise history of early targeting and force planning ideas, see Desmond Ball, “The Development of the SIOP, 1960–1983,” in Desmond Ball and Jeffrey Richelson, Strategic Nuclear Targeting (Ithaca, N.Y.: Cornell University Press, 1986), pp. 57–83.

104. See Bruce G. Blair, Strategic Command and Control: Redefining the Nuclear Threat (Washington, D.C.: Brookings, 1985); and Bruce G. Blair, Global Zero Alert for Nuclear Forces (Washington, D.C.: Brookings, 1995).

105. See Scott D. Sagan, The Limits of Safety: Organizations, Accidents, and Nuclear Weapons (Princeton, N.J.: Princeton University Press, 1993).

106. See Barry R. Posen, Inadvertent Escalation: Conventional War and Nuclear Risks (Ithaca, N.Y.: Cornell University Press, 1991).

107. Matthew Bunn and Kosta Tsipis, “The Uncertainties of a Preemptive Nuclear Attack,” Scientific American (November 1983).

108. David Mosher, “Appendix B: Exchange Calculations,” in Raymond Hall, David Mosher, and Michael O’Hanlon, The START Treaty and Beyond (Washington, D.C.: Congressional Budget Office, 1991), pp. 143–65. Other examples include the Soviet SS-19 with a lethal radius of 300 meters, the U.S. Minuteman IIIA with a lethal radius of 185 meters, and the U.S. D5/Mark 5 (Trident II SLBM) warhead with a lethal radius of 210 meters.

The lethal radius of a warhead is calculated by taking the one-third power of (Y/H), where Y is the attacking warhead yield measured in megatons and H the hardness of the attacked missile silo in thousands of pounds per square inch, and then multiplying that by 460 meters. So a two megaton warhead attacking a 2,000 pound/square inch silo would have a lethal radius of 460 meters.

The single-shot survival probability for a silo attacked by a given warhead is then calculated as follows. First, square the lethal radius of the warhead (that is, take it to the second power, or multiply it by itself). Then, square the circular error probable of the warhead, and multiply the resulting number by 1.44, and then make it negative. Take the ratio of the first term (the square of the lethal radius) to the second (the negative of 1.44 times the square of the circular error probable). That overall expression is then the power to which the naturally occurring number “e” is taken. In short, Single Shot Survival = [exp][-LRsquared/1.44CEPsquared], where LR = lethal radius and CEP = circular error probable.

109. Samuel Glasstone, ed., The effects of Nuclear Weapons, revised edition (Washington, D.C.: Government Printing Office, 1962), pp. 134–35, 156–76; Mosher, “Appendix B: Exchange Calculations,” in Hall, Mosher, and O’Hanlon, The START Treaty and Beyond, p. 159; and David Ochmanek and Lowell H. Schwartz, The Challenge of Nuclear-Armed Regional Adversaries (Santa Monica, Calif.: RAND, 2008), p. 7. Bombers would likely be damaged once struck by two to four pounds per square inch of overpressure or more.

110. See Sumit Ganguly, Conflict Unending: India–Pakistan Tensions Since 1947 (New York: Columbia University Press, 2001).

111. See Stephen Philip Cohen, The Idea of Pakistan (Washington, D.C.: Brookings, 2004), pp. 97–130.

112. See International Crisis Group, Unfulfilled Promises: Pakistan’s Failure to Tackle Extremism (Brussels, 2004).

113. General David H. Petraeus, Lt. General James F. Amos, and Lt. Colonel John A. Nagl, The U.S. Army/Marine Corps Counterinsurgency Field Manual (Chicago: University of Chicago Press, 2007), p. 23.

114. James Dobbins, John G. McGinn, Keith Crane, Seth G. Jones, Rollie Lal, Andrew Rathmell, Rachel Swanger, and Anga Timilsina, America’s Role in Nation Building from Germany to Iraq (Santa Monica, Calif.: RAND, 2003), p. xvii.

115. Michael E. O’Hanlon, Saving Lives with Force (Washington, D.C.: Brookings, 1997), pp. 38–42; and James T. Quinlivan, “Force Requirements in Stability Operations,” Parameters, vol. 25, n. 4 (Winter 1995–1996), pp. 59–69.

116. Seth G. Jones, Jeremy M. Wilson, Andrew Rathmell, and K. Jack Riley, Establishing Law and Order After Conflict (Santa Monica, Calif.: RAND, 2005), p. 19; James Dobbins, John G. McGinn, Keith Crane, Seth G. Jones, Rollie Lal, Andrew Rathmell, Rachel Swanger, and Anga Timilsina, America’s Role in Nation-Building: From Germany to Iraq (Santa Monica, Calif.: RAND, 2003); and James Dobbins, Seth G. Jones, Keith Crane, Andrew Rathmell, Brett Steele, Richard Teltschik, and Anga Timilsina, The UN’s Role in Nation-Building: From the Congo to Iraq (Santa Monica, Calif.: RAND, 2005).

117. See Center on International Cooperation, Annual Review of Global Peace Operations, 2007 (Boulder, Colo.: Lynne Rienner Publishers, 2007), p. 3; International Institute for Strategic Studies, The Military Balance 2008 (Oxfordshire, England: Routledge, 2008), p. 448, plus enclosed map; and Michael E. O’Hanlon, Expanding Global Military Capacity for Humanitarian Intervention (Washington, D.C.: Brookings, 2003), pp. 56–57.

118. International Institute for Strategic Studies, The Military Balance 2003–2004, pp. 140–42.

119. Angel Rabasa, Lesley Anne Warner, Peter Chalk, Ivan Khilko, and Paraag Shukla, Money in the Bank: Lessons Learned from Past Counter-insurgency (COIN) Operations (Santa Monica, Calif.: RAND, 2007), pp. ix–xv, 1–4.

120. Andrew F. Krepinevich, Jr., The Army and Vietnam (Baltimore, Md.: Johns Hopkins Press, 1986), pp. 177–214; and Robert S. McNamara, In Retrospect: The Tragedy and Lessons of Vietnam (New York: Vintage Books, 1995), pp. 169–77, 210–12, 220–23, 233–47, 262–63, 282–93.

121. Neil Sheehan, A Bright Shining Lie (New York: Vintage Books, 1988), pp. 201–65.

122. David Galula, Counterinsurgency Warfare: Theory and Practice (New York: Praeger, 2005), pp. 70–86.

123. Krepinevich, The Army and Vietnam, pp. 172–77.

124. General David H. Petraeus, Lt. General James F. Amos, and Lt. Colonel John A. Nagl, The U.S. Army/Marine Corps Counterinsurgency Field Manual (Chicago: University of Chicago Press, 2007), pp. 1–52; Steven Metz, Learning from Iraq: Counterinsurgency in American Strategy (Carlisle, Pa.: Army War College Strategic Studies Institute, 2007), pp. 1–30; and Thomas E. Ricks, Fiasco: The American Military Adventure in Iraq (New York: Penguin Press, 2006), pp. 149–202.

125. International Institute for Strategic Studies, The Military Balance 2003–2004, pp. 136–37, 337.

126. General David H. Petraeus, Lt. General James F. Amos, and Lt. Colonel John A. Nagl, The U.S. Army/Marine Corps Counterinsurgency Field Manual (Chicago: University of Chicago Press, 2007), p. 23.

127. Michael E. O’Hanlon, Saving Lives with Force (Washington, D.C.: Brookings, 1997), pp. 38–42; and James T. Quinlivan, “Force Requirements in Stability Operations,” Parameters, vol. 25, n. 4 (Winter 1995–1996), pp. 59–69.

128. On Indonesia, see Robert Karniol, “Country Briefing: Indonesia,” Jane’s Defence Weekly, April 7, 2004, pp. 47–52.

129. Krepinevich, The Conflict Environment of 2016, pp. 23–27.

130. In theory, calculating ADE totals from equipment inventories and force structure tables (as listed in available databases) is quite difficult and tedious. In practice, a good approximation can be obtained by beginning with active equipment inventories, starting with main battle tanks. For example, India has about 4,000 in its active stocks, featuring 330 T-90 and almost 2,000 T-72 and about 1,000 indigenous Vijayanta tanks (as well as some 700 T-55s being gradually replaced by T-90s). Typically, there are 300 main battle tanks in a heavy division, so these inventories are enough for thirteen divisions in principle–though perhaps just eleven to twelve in practice assuming a modest reserve of excess tanks to replace those in depot or otherwise temporarily unserviceable. All of these tanks except the T-55 are reasonably modern and within 10 to 20 percent of the “score” of an M1 tank (following the conservative methodology of the WEI-WUV system). The T-55 stocks by contrast are much inferior, worth at most half what a modern tank might be in terms of quality. So compensating for this fact, and the attrition/maintenance reserve mentioned here, India might have the equivalent often ADEs of tank inventory. See International Institute for Strategic Studies, The Military Balance 2008 (Oxfordshire, England: Routledge, 2008), pp. 341–51.

131. Christopher Bowie, Fred Frostic, Kevin Lewis, John Lund, David Ochmanek, and Philip Propper, The New Calculus: Analyzing Airpower’s Changing Role in Joint Theater Campaigns (Santa Monica, Calif.: RAND, 1993), pp. xi–xxii.

132. See David A. Ochmanek, Edward R. Harshberger, David E. Thaler, and Glenn A. Kent, To Find, and Not to Yield: How Advances in Information and Firepower Can Transform Theater Warfare (Santa Barbara, Calif.: RAND, 1998), pp. 6–12, 31–42, 77–85.

133. Trevor N. Dupuy, Attrition: Forecasting Battle Casualties and Equipment Losses in Modern War (Fairfax, Va.: HERO Books, 1990), p. 151.

134. Krepinevich, The Conflict Environment of 2016: A Scenario-Based Approach (Washington, D.C.: Center for Strategic and Budgetary Assessments, 1996), pp. 11–15; and Caitlin Talmadge, “Closing Time: Assessing the Iranian Threat to the Strait of Hormuz,” International Security, vol. 33, no. 1 (Summer 2008), pp. 82–117.

135. International Institute for Strategic Studies, The Military Balance 2008, p. 243.

136. The formula for radar horizon is RH = square root of (diameter of Earth × altitude of satellite). This formula follows directly from the Pythagorean theorem, drawing a right triangle with one side the radius of the Earth, a second side the distance from the satellite in question to the farthest point on Earth’s surface within its view, and a third side from the center of the Earth to the satellite (this latter segment is the triangle’s hypotenuse). Symbolically, RH = √(DA). Since the diameter of the Earth is about 8,000 miles, an aircraft at just under eight miles’ altitude can see about 250 miles.

KEY REFERENCES AND SUGGESTIONS
FOR FURTHER READING

Betts, Richard K., Surprise Attack (Washington, D.C.: Brookings, 1982).

Biddle, Stephen, Military Power: Explaining Victory and Defeat in Modern Battle (Princeton, N.J.: Princeton University Press, 2004).

Blair, Bruce G., Strategic Command and Control: Redefining the Nuclear Threat (Washington, D.C.: Brookings, 1985).

Bowie, Christopher, Fred Frostic, Kevin Lewis, John Lund, David Ochmanek, and Philip Propper, The New Calculus (Santa Monica, Calif.: RAND, 1993).

Bunn, Matthew, and Kosta Tsipis, “The Uncertainties of a Preemptive Nuclear Attack,” Scientific American (November 1983).

Center on International Cooperation, Annual Review of Global Peace Operations, 2007 (Boulder, Colo.: Lynne Rienner Publishers, 2007).

Cordesman, Anthony H., and Abraham R. Wagner, Lessons of Modern War, Volume 2: The Iran–Iraq War (Boulder, Colo.: Westview Press, 1990).

Cote, Owen and Harvey Sapolsky, Antisubmarine Warfare After the Cold War (Cambridge, Mass.: MIT Security Studies Program, 1997).

Dobbins, James, John G. McGinn, Keith Crane, Seth G. Jones, Rollie Lal, Andrew Rathmell, Rachel Swanger, and Anga Timilsina, America’s Role in Nation Building from Germany to Iraq (Santa Monica, Calif.: RAND, 2003).

Dunnigan, James F., How to Make War: A Comprehensive Guide to Modern Warfare for the Post–Cold War Era (New York: William Morrow and Company, Inc., 1993).

Dupuy, Trevor N., Numbers, Predictions, and War: The Use of History to Evaluate and Predict the Outcome of Armed Conflict, revised edition (Fairfax, Va.: HERO Books, 1985).

Epstein, Joshua M., Strategy and Force Planning: The Case of the Persian Gulf (Washington, D.C.: Brookings Institution, 1987).

General Accounting Office (now the Government Accountability Office), Operation Desert Storm: Evaluation of the Air Campaign (Washington, D.C.: GAO, June 1997), GAO/NSIAD-97-134.

Hoeber, Francis P., Military Applications of Modeling: Selected Case Studies (New York: Gordon and Breach Science Publishers, 1981).

Hughes, Capt. Wayne P., Jr. (U.S. Navy, retired), Fleet Tactics and Coastal Combat, 2^nd edition (Annapolis, Mary land: Naval Institute Press, 2000).

Johnson, Stuart E. and Duncan Long, eds., Coping with the Dragon: Essays on PLA Transformation and the U.S. Military (Washington, D.C.: National Defense University, December 2007).

Kamphausen, Roy and Andrew Scobell, eds., Right-Sizing the People’s Liberation Army: Exploring the Contours of China’s Military (Carlisle, Pa.: Strategic Studies Institute, Army War College, 2007).

Keaney, Thomas A. and Eliot A. Cohen, Gulf War Air Power Survey Summary Report (Washington, D.C.: Government Printing Office, 1993).

Krepinevich, Andrew F., The Conflict Environment of 2016 (Washington, D.C.: Center for Strategic and Budgetary Assessments, 1996).

Lambeth, Benjamin S., NATO’s Air War for Kosovo: A Strategic and Operational Assessment (Santa Monica, Calif.: RAND, 2001).

Miller, Steven E., ed., Conventional Forces and American Defense Policy (Princeton, N.J.: Princeton University Press, 1986).

Mulvenon, James C., Murray Scot Tanner, Michael S. Chase, David Frelinger, David C. Gompert, Martin C. Libicki, and Kevin L. Pollpeter, Chinese Responses to U.S. Military Transformation and Implications for the Department of Defense (Santa Monica, Calif.: RAND Corporation, 2006).

Petraeus, General David H., Lt. General James F. Amos, and Lt. Colonel John A. Nagl, The U.S. Army/Marine Corps Counterinsurgency Field Manual (Chicago: University of Chicago Press, 2007).

Posen, Barry R., Inadvertent Escalation: Conventional War and Nuclear Risks (Ithaca, N.Y.: Cornell University Press, 1991).