Military professionals often seek a "scientific" understanding of war. This approach is appealing because the human mind tends to organize its perceptions according to familiar analogies and metaphors, like the powerful images of traditional Newtonian physics. Such metaphors can be very useful. Our military doctrine abounds with terms like "leverage," "center of gravity," and "mass."

Useful as it is, attempting to apply such a "scientific" approach can result in some very unrealistic ideas. For example, one widely read military historian recently tried to reduce military power to a simple and, he argued, reliably predictive scientific equation: P=NVQ (where P = combat power, N = numbers of troops, V = "variable factors," and Q = quality of troops).*9 In practice, of course, strategists must seek some understanding of all of these factors. However, even an accurate figure for N is surprisingly difficult to find, while V and Q are impossible to reliably quantify—or even to define. This kind of mathematical approach, even though it reflects some important truths, cannot serve as a practical basis for strategic analysis and prediction.

Similarly, many political scientists treat political entities as "unitary rational actors," the social equivalents of Newton's solid bodies hurtling through space. Real political units, however, are not unitary. Rather, they are collections of intertwined but fundamentally distinct actors and systems. Their behavior derives from the internal interplay of both rational and irrational forces, as well as from the peculiarities of their own histories and of sheer chance. Strategists who accept the unitary rational actor model as a description of entities at war will never understand either side's motivations or actual behavior. Such strategists ignore their own side's greatest potential vulnerabilities and deny themselves potential levers and targets—the fault-lines that exist within any human political construct. In fact, treating an enemy entity as a unitary actor tends to be a self-fulfilling and counterproductive prophecy, reinforcing a sense of unity among disparate elements which might otherwise be pried apart.

Fortunately, the physical sciences have begun to embrace the class of problems posed by social interactions like human politics and war. Therefore, "hard-science" metaphors for war and politics can still be useful. The appropriate imagery, however, is not that of Newtonian physics. Rather, we need to think in terms of biology, and particularly ecology.

To survive over time, the various participants in any ecosystem must adapt—not only to the "external" environment but to each other. These agents compete or cooperate, consuming and being consumed, joining and dividing, and so on. In fact, from the standpoint of any individual agent, the behavior of the other agents is itself a major element of the environment. The collective behavior of the various agents can even change the nature of the "external" environment. For instance, certain species, left unchecked, can turn a well-vegetated area into a desert. Such changes in the environment will, in turn, necessitate and reward adaptive changes elsewhere in the system. And, of course, the environment can also be changed by the intrusion of external factors, setting off yet another round of adaptations.

A system created by such a multiplicity of internal feedback loops is called a complex adaptive system. Such systems nestle one inside the other, constructing, interpenetrating, and disrupting one another across illusory "system boundaries." Any individual member of a plant or animal species, for example, is a complex adaptive system made up of cells. Its protective skin or shell encloses an environment quite different from that outside the system the cells collectively have created. That individual, in turn, is part of larger complex adaptive systems. It is part of a local breeding population that is part of a species, both of which have an existence above and beyond the individual organism. Both the individual and the species it is part of belong to another order of complex adaptive system, the local ecosystem. And so on.

Such systems are inherently dynamic. Although they may sometimes appear stable for lengthy periods, the complex network of interconnected feedback loops demands that its subcomponents constantly adapt or fail. No species evolves alone; rather, each species "co-evolves" with the other species that make up its environment. The mutation or extinction of one species in any ecosystem will have a domino or ripple effect throughout the system, threatening damage to some species and creating opportunities for others. Slight changes are sometimes absorbed unnoticed by the system. Other slight changes—an alteration in the external environment or a local mutation—can send the system into convulsions of growth or collapse. Sometimes both simultaneously.

One of the most interesting things about complex adaptive systems is that they are inherently unpredictable. It is impossible, for example, to know in advance which slight perturbations in an ecological system will settle out unnoticed and which will spark catastrophic change. This is so, not because of any flaw in our understanding of such systems, but because the system's behavior is generated according to rules the system itself develops and is able to alter. In other words, a system's behavior may be constrained by external factors or laws, but is not determined by them. Every system evolves according not only to general laws but to local rules established by evolution, accident, and happenstance—or, if an intelligent agent is involved, through conscious intervention.

For all of these reasons, systems starting from a similar base will come to have unique individual characteristics based on their specific histories. Science can describe and often explain the behavior and evolution of a complex adaptive system, but cannot predict it. Oftentimes, however, the chain of events is so subtle and convoluted, and the evidence so fragmentary, that the sequence of events and the web of causation can never be satisfactorily understood, even in retrospect.

The reason we dwell on the complex adaptive system is that it provides so much insight into human political constructs. Any group of humans who interact will, over time, form a unique system broadly similar to the ones we have described. Humans build all sorts of social structures and engage in complex behavior. Human structures include families, tribes, clans, social classes and castes, secret societies, street gangs, armies, feudal hierarchies, commercial corporations, church congregations, political parties, bureaucracies, criminal mafias, states of various kinds, alliances, confederations, and empires. These structures participate in separate but thoroughly intertwined networks we call social, economic, and political systems. Those networks produce markets, elections, and wars. And, though we tend to associate war with the state, in fact all of the structures we have listed have been known to engage in warfare.

Such networks and structures create their own rules, and are thus fundamentally unpredictable. Economic and political events can be subjected to rigorous analysis before and while they occur, and can be described and often plausibly explained afterwards. Nonetheless, as any regular watcher of the evening news has long since discovered, they cannot reliably be predicted. Indeed, both evolutionary scientists and historians of human events find steady employment in seeking better ways to "postdict" the past, which can be just as puzzling as the present or future. We can certainly see "patterns" in human history, yet history does not repeat itself. "Victory" goes, not only to those participants who learn the existing rules, but also to those who succeed in making new ones.

When we say that politics and war are unpredictable, we do not mean that they are sheer confusion, without any semblance of order. Intelligent, experienced military and political actors are generally able to foresee the probable near-term results, or at least a range of possible results, of any particular action they may take. Broad causes, such as a massive superiority in manpower, technology, economic resources, and military skill will definitely influence the probabilities of certain outcomes. Conscious actions, however, like evolutionary adaptations, seldom have only their intended effects. As many political scientists and historians have wryly observed, there is an unremitting "law of unintended consequences." As the ripples from any one action spread out, their effects unpredictably magnify or nullify the ripples from other actions. Thus events that seemed at the time to have great importance may prove to lead nowhere, while actions so minor as to escape notice may have tremendous consequences. Further, human systems are "open" systems, without any clear boundaries. Events wholly outside the range of political and military leaders' vision can have an unforeseen impact on the situation. New economic and social ideas, technological innovations with no obvious military applications, changes in climatic conditions, demographic shifts, all can lead to dramatic political and military changes. Enemy actions, friction, imperfect knowledge, low order probabilities, and sheer random chance introduce new variables into any evolving situation. Events begin to spin out of control. History is too full of examples of great states defeated by seemingly inferior powers, of experienced leaders and armies overthrown by inexperienced newcomers, to believe that politics and war are predictable, controllable phenomena.

Thus it is seldom enough to set a good strategic course and follow it through. As the great German military leader Helmuth von Moltke said, "No plan survives first contact with the enemy." Effective strategists must have a feel for the nature of this environment and be prepared for both the unexpected setbacks and the sudden opportunities it is certain to deliver. Military strategy demands a capacity for both painstaking planning and energetic adaptation to unfolding events.