The first attempt to make computers “talk” to each other took place in 1969. Once the Net had become widely available, computers rapidly took the place of messengers, telephones, and other devices in facilitating communication among participants. This often caused the premises where the games were held to look like old-fashioned telephone exchanges with cables snaking about in all directions. Nor did the Net stay limited to its inventors, i.e. the US military. Instead it spread to the civilian world, eventually making possible well-known amateur games such as World of Warcraft and Warhammer in which hundreds of players can play simultaneously. Last but not least, initially computers were few and far between. Down to the mid 1980s this scarcity provided the fields to which they were applied, wargaming specifically included, with an aura of mystery and sophistication. As Northcote Parkinson of Parkinson’s Law fame once noted, that was not the least of their merits.5 To move around little counters or miniatures on a board, or map, or the floor of a room, or in a sandbox, and pretend to kill them as H. G. Wells and countless other boys aged two to a hundred and fifty did was one thing. To enlist a computer costing millions and functioning in a way few understood in order to do exactly the same was a different and much more respectable one.
Not only did computers assist human wargamers but they could, up to a point, take their place. One way of doing this was by using artificial intelligence, meaning a program which, in a certain sense, can “understand” moves and respond to them according to a pre-programmed set of rules. As long as the game is sufficiently simple and structured, indeed, an artificial intelligence program may even “learn” to tailor its moves to the kind of opponent it is playing against, adjusting its play according to whether he is aggressive, defensive, and so on. The potential of this approach has been demonstrated by what in some ways is the most important wargame of all, i.e. chess. The pioneers of chess-playing computer programs were mathematicians such as Norbert Wiener, Claude Shannon, and Alan Turing who did their work during the late 1940s and early 1950s. In 1967 a program called Mac Hac for the first time defeated a human in tournament play. Such programs have been brought to the point where, since 2000 or so, they have been regularly defeating even world champions.6 The magnitude of the achievement may be appreciated by the fact that, in an average game consisting of a few dozen moves, the number of possible positions is in the order of 1×1040.
The second and in some ways even more radical approach is to dispense with the human player altogether. Games, or perhaps it would be better to call them simulations, can then be run by having two (or more) programs “fight” each other inside the computer itself. Simply juggling the numbers that go into the equations which make up the programs they run will produce different outcomes. Suppose a simulation whose purpose is to help select the best combination of weapons for carrying out such and such a task. Initially each side is given so and so many weapons of such and such a kind. They carry so and so much ammunition and are effective at such and such a range. Given such and such conditions they secure such and such a percentage of hits, inflicting such and such damage on such and such targets defended by such and such means. Changing the numbers will change the results. Hopefully multiple repetitions will enable us to discover the best combination: namely, the one that will produce the greatest number of enemy casualties within the shortest time at the least cost.
Most if not all simulations will be far more complex than this example suggests. Still, the underlying principle is the same. The great advantage of this approach is the fact that, once the equations have been written and the program put in place (once the model has been created) the game, or simulation as some prefer to call it, can be repeated as often as one likes, very fast, and at almost no additional cost. That is not true of humans, who, even if they can be induced to play a game for a second time and even if the cost of doing so is acceptable, will inevitably enter the second game with a set of ideas and assumptions different from the ones with which they entered the first. Thus the ancient Greek maxim that one cannot enter the same river twice applies to wargames as it does to “real” life – an interesting observation in itself.
The benefits of computers were not evenly distributed among various kinds of wargames. One end of the spectrum was occupied by games seeking to simulate not merely war but politics, of which the above-mentioned Sigma Series and Global Wargames were prime examples. To be sure, almost everybody nowadays pays at least lip service to the critical role that politics plays, or ought to play, in initiating war, governing its conduct, and terminating it. However, its imprecise and subjective nature makes developing any kind of rule to model its conduct next to impossible. This specifically includes deterrence, both nuclear and conventional, a “squishy” subject if ever there was one and one that has a lot more to do with psychology than with any branch of natural science.7 Supposedly game theory is available to help cope with the problem, providing optimal solutions to precisely formulated problems. However, so complex is reality that only rarely can the theory be usefully applied to it. Very often even the most elementary terms are the subject of disagreement among designers.
To illustrate the last-mentioned point let us return to Robert Aumann, surely as qualified an expert on game theory as is alive today. An Israeli citizen, politically Aumann is located on the religious far right. He has criticized the decision by former prime minister Ariel Sharon to evacuate Gaza as “idiotic.”8 In one of his articles he refers to a well-known game named “blackmailer’s paradox.” In it an insane person, let us call him Brian, forces a sane one, let us call him Alex, to adopt an irrational line of behavior.9 The logic as developed in the game is impeccable. What is anything but impeccable is Aumann’s attempt to fit the Arab−Israeli conflict into the game, or perhaps the game onto the Arab−Israeli conflict, by claiming that “the Arabs present rigid and unreasonable opening positions at every negotiation,” blackmailing their opponents.10 He thus makes a value judgment that not even many Israelis, let alone a single Arab, would accept. The difficulty of agreeing what constitutes blackmail, or, turning back to Brian and Alex, which one of the two of them is insane, explains why, in games designed to simulate politics, computers are largely restricted to performing auxiliary tasks.
At the opposite end of the scale from politics, again specifically including deterrence, stands strategic nuclear warfare. By that is meant the kind of nuclear warfare that is directed against the opponent’s homeland. Without any doubt, such warfare is by far the most deadly and destructive form of armed conflict ever invented. So destructive, in fact, that whereas previous weapons only helped change the ways war is waged, nuclear ones seem to have altered the reasons why it is waged as well as the objectives it may be waged for.11 To use Churchill’s phrase, increasing the power of existing warheads would merely make the rubble bounce. Hence not only is this situation likely to prevail in the future too, but there seems to be little incentive for developing such doomsday methods as “earthquake warfare” and “tsunami warfare.”
That much taken for granted, there are several reasons why nuclear warfare is relatively easy to game. First, the number of weapons and delivery vehicles is much smaller than in any other form of war – under certain scenarios perhaps no more than one. Second, owing to the enormous power of the weapons and their residual effects, such as radioactive fallout, it can only be waged far away from friendly troops and populations. Conversely, exercises held as early as the mid 1950s showed that, if it was waged anywhere near those troops and those populations, the outcome might well be to end the existence of both.12 This requirement in turn means that warheads must be delivered at long distance either through the air or through outer space. However, air and space present much simpler mediums than the sea, let alone the land. While some meteorological phenomena do apply, there are no mountains, lakes, rivers, swamps, forests, various kinds of transportation arteries, villages and towns that must be taken into account in planning one’s moves and carrying them out. By comparison, calculating th
e flight paths of aircraft and the trajectories of missiles is easy.
Third, this very simplicity means that the man/machine balance is markedly different. Starting early in the nineteenth century, the Industrial Revolution caused warfare to become much more capital-intensive than it was before. In this respect, as in so many others, the advent of nuclear weapons marked a watershed. The outstanding characteristic of nuclear warfare is precisely that it is waged with the aid of comparatively few humans. Piloting or launching or otherwise controlling enormously expensive machines, this handful produces a vast bang: as has been well said, instead of equipping the man we now man the equipment.13 Extremely rigid procedures are put into place so as to exercise complete control and prevent unauthorized or accidental war.14 Coupled with secure, hardened, redundant communications these procedures are designed to minimize friction of any kind. Last but not least, machines are completely artificial creations. Though their performance may not be entirely error-free, it is far better understood and far more predictable than that of variable, capricious humans. Which means that it is also much easier to model and game.
Even the relatively few humans involved in such warfare are of a very special kind. When it comes to launching missiles, or providing warning against them, or defending against them, most of what Clausewitz calls the Strapazen – meaning, strain, fatigue, or suffering – so critical in other kinds of war are all but absent. To be sure, guiding a heavy strategic bomber to its target is not the easiest job in the world. On the other hand, whatever problems they may face, pilots are neither hungry nor thirsty, nor driven half crazy by long periods of enforced sexual abstinence. This is even more true of drone and missile operators who are not expected to face danger or physical discomfort of any kind. Though the stress may be enormous, it is of the kind experienced by air traffic controllers rather than by combat soldiers, who not only operate under what are often extremely difficult circumstances but risk their lives as well. The operators are stationed in specially constructed facilities, possibly located thousands of miles from the scene of action. Everything is done to insulate them and make them comfortable so they can focus on their work. They sit in front of screens, watch blips, talk to other operators, and manipulate controls. In some cases all they are required to do is receive orders that tell them to insert a key into a slot and turn it. This, incidentally, explains why the percentage of women among them is especially high.15
During the early Cold War years the Soviet Union was, to quote Churchill’s famous expression, “a riddle wrapped in a mystery inside an enigma.” To some extent this remained true right down to the final collapse of the Soviet Union; thanks to the development of reconnaissance satellites, though, major parts of the mystery began to be lifted from the early 1960s on. The number of bomber aircraft, missiles, and, later, cruise missiles on each side was roughly known. So, in many cases, were the locations of the bases where they were stationed, their technical characteristics (such as the ability to take off or be launched quickly before a strike by the other side would neutralize them), their failure rate, their range, the distance at which they would be discovered by the other side’s radar, the time they would take to reach their targets, their size and shape (which could be correlated roughly with the number of warheads Soviet aircraft and missiles carried), their accuracy, and the size of the warheads they carried. Those factors in turn could be correlated with population densities, the ability of different kinds of structures to resist blast, and the like. Assuming such and such a target, within such and such a distance from ground zero, so and so many people would be killed and such and such a percentage of buildings demolished. Assuming so and so much radioactivity and winds blowing from such and such a direction, so and so many people could expect to die of leukemia when they were reached by radioactive fallout.16
To a lesser but still very considerable extent, the same also applied to whatever defenses had been developed to deal with bombers and missiles. Even specific strategies such as first strike, second strike, and the like could be simulated without much difficulty.17 To be sure, some things, notably the way each side might interpret the other’s preparations and moves and react to them, remained unknown. For example, would a doctrine aimed at targeting the other’s delivery vehicles while leaving his cities alone induce the opponent to respond in kind, thus sparing the world from utter destruction in case a war broke out? Or would it merely provide him with an incentive to switch to “launch on warning” (rather than wait for the first hostile missiles to hit his soil) and strike first, thus endangering the balance of terror and making war more likely?18 Still, such games, and the mathematics that made them possible, were extremely useful in answering questions as to which kind of delivery vehicles, aircraft or ballistic missiles, were preferable; which of one’s own forces to use first and which ones to keep in reserve; and so on. Applying similar methods to the other side’s armed forces, one could turn the game around and obtain an estimate of one’s own losses and how to minimize them, as in choosing between providing the population with nuclear shelters and building an anti-ballistic missile defense system, or in deciding just how such a system should be constructed and organized. All in all, probably never before had such a large part of war been reducible to the laws of physics.
If gaming nuclear warfare is relatively easy it is also exceptionally important. Throughout history, commanders at every level have always based their plans for the future on the experience of the past. Especially from the last decades of the nineteenth century on, one of the most important instruments many of them used for the purpose were wargames. In the case of nuclear warfare, though, no such experience was available. Speculation and science fiction aside, this fact left wargames as almost the only available tool for understanding it, planning it, and testing it, vastly enhancing the role that they could and should play. Probably for the first time in history BYMs (bright young men) with no military experience, some of them just out of college, could plausibly claim to know as much about future warfare – or at any rate the most powerful and most destructive forms of future warfare – as the most senior military commanders of the time. Some actually boasted of never having heard a shot fired in anger.
The BYMs’ self-imposed mission was nothing less than to invent entirely new techniques for dealing with what they saw, not without reason, as entirely new military problems. In doing so, they sought to combine imagination with rigid quantification. It hardly requires saying that this situation led to numerous clashes between different generations and different cultures: after all, few organizations resemble each other less than the military on the one hand and think tanks on the other. For example General Lemay, during his term as commander of the Strategic Air Command (1949–57), was quoted as saying that, at the time when he himself was directing hundreds of heavy bombers against Japan and dropping atomic bombs on Hiroshima and Nagasaki, Harold Brown, then a young analyst and later Secretary of Defense under President Carter, had been in junior high school. On the other side Herman Kahn, answering critics who questioned his methods, used to ask when they had last fought a thermonuclear war.19 As the saying goes, old views do not die; it is those who hold them who do. Under McNamara during the 1960s the BYMs’ victory seemed complete – until, that is, the Vietnam War caused a reaction to set in. But that is another story.
All this explains why, inside the so-called defense community, thousands if not tens of thousands of nuclear wargames have been held over the years. Most of the details are understandably secret, and at any self-respecting war college the building where the games are held is always the most tightly guarded of all. Still, we know that the Joint Chiefs of Staff, preparing the first Strategic Integrated Operations Plans (SIOPs) in 1961–2, used wargames to determine the level of damage that the US and Soviet Union might inflict on one another in an all-out exchange of nuclear strikes. Not long afterwards, the chief of the Joint Chiefs of Staff, General Lyman Lemnitzer, presented the results to President Kennedy.20 Lemnitzer never me
ntioned the possibility of an “out of the blue” US offensive. What he did tell the president was that the number of US casualties would depend primarily on the success of the air force in eliminating as many Soviet bombers and missiles as possible before they could be launched. Yet almost certainly not even the best planned attack, perfectly executed, would prevent some of the Kremlin’s nuclear forces from “riding out” the attack – as the phrase went – and retaliating in kind. The outcome would be horrendous casualties, even though their exact number was not spelt out in the briefing.
As was noted at the time and later, the SIOP in question was extremely rigid. The objective was to deter, and if necessary safeguard against, a Soviet first strike by using the available forces before they could be knocked out. Hence the plan, and presumably the games on which it was partly based, only provided for all-out nuclear warfare and very little else. Technological progress, specifically the growing number of intercontinental ballistic missiles (ICBMs) and the introduction from about 1965 on of Multiple Independent Reentry Vehicles or MIRVs, led to changes in this respect. ICBMs in their silos were much harder to hit and destroy than bombers on their airfields, thus easing the awesome dilemma of “use them or lose them” and providing their owners with some flexibility in deciding what to do. At the same time, arming ballistic missiles with multiple reentry vehicles capable of being independently guided not only increased the number of targets each of them could hit but also enabled them to do so much more accurately. They could now be aimed not merely at cities but at much smaller targets, which in turn led to the development of smaller warheads. The same applied to the cruise missiles that entered the arsenal from the 1970s on.21 Escalation, in other words, no longer looked automatic, nor the destruction of the world inevitable.
Wargames Page 38