7.2 REINSTATING THE PRIMACY OF SYSTEMS ANALYSIS
The scenario-independent process described above would provide a perfectly acceptable way of controlling command technology provided we found a way of reconstituting the deliberative method of systems analysis which the technique of modeling and simulation has arrogantly brushed aside. Without such a method, the modeling and simulation that unfortunately passes for analysis today, will not help decision makers judge which of the many alternative technologies made possible by command technology would best deal with a particular vulnerability.
Indeed, to help the decision maker’s judgment, one needs to speak in terms that reflect the issues confronting him. These issues can only be captured by, what we have called in Chapter 3, the critical factors associated with the decision; in fact, the only place where deliberation can take a foothold is precisely where these critical factors come in. However, because the critical factors are not components of a war machine or instances in the evolution of a war, one cannot simulate them and, therefore, modeling and simulation does not even attempt to identify such factors. A simulation sheds no light upon any of the issues that a decision maker must weigh on his way to a decision, because all the critical factors that need weighing are hidden behind a set of numbers describing the final outcome of the war. Consequently, modeling and simulation will never be able to contribute to the debate about future technology. By contrast, systems analysis illuminates the debate about technological choices precisely because it seeks to understand what critical factors underlie that debate and develops its models accordingly.
Performing a large number of parametric analyses to explore the sensitivity of a simulation’s outcome to changes in the many input parameters that drive the simulation will not help illuminate the issues involved in the decision, despite all the protestations to the contrary that are made by the practitioners of the modeling and simulation technique. First, parametric analysis varies the input parameters, not the laws that are supposed to govern the operations modeled or, for that matter, the relationships between such operations. Second, by virtue of the very large number of input parameters that generally drive such simulations, varying them all would generate far more information than human understanding can intelligently handle; one gets a lot of curves, but no comprehensive understanding of the message they might convey with respect to the decision at hand. Finally, and this might be more important to decision making than the previous two considerations, one cannot test the effect of the fundamental assumptions involved in the simulation upon its results because one has made no effort to discover what those assumption actually were.
Under these circumstances, having discovered how to replace the Cold War scenario with a search for singular vulnerabilities will have bought us nothing; at the end of the day, the decision maker will have to render a decision concerning his choice of technology, whether that decision is geared to winning in a particular scenario or to eliminating a particular vulnerability, without any substantive input from the analytic community. The fact that sponsors in the Pentagon nevertheless continue to ask for studies in support of their deliberations is more a consequence of historical inertia and bureaucratic process than a tribute to the value they expect to obtain from it.
While, as we have seen above, dealing with the scenario problem required more invention than will, returning the technique of modeling and simulation to the subservient position it once held with respect to the deliberative method of systems analysis, requires more will than invention. One need not invent a new method of deliberation since the method of systems analysis, together with the techniques described in Chapter 3, should serve the new principle of eliminating vulnerabilities every bit as well as it did the old principle of executing the planning scenario. What is needed instead is a renewed dedication to the proposition that technique must be guided by method, and that the method should connect to the deliberative process it tries to illuminate. Before the memory of the heydays of systems analysis will have completely faded from memory, an effort must be made to record and proselytize its method as a guide to the young generation entering the field of defense analysis.
Unfortunately, the guidance young employees receive when they enter the field, points them towards computer simulation and database management, not towards understanding the issues involved in the decision. Any notion that their professional advancement would benefit from trying to think a problem through in advance of modeling it, is often firmly discouraged. People being what they are, these disincentives, particularly when coupled with the lack of adequate educational literature on the subject and the virtual absence of any support structure intended to pass experience from the old to the new generation, are designed to maintain the current supremacy of modeling and simulation over defense analysis for the foreseeable future.
Given this absence of focus upon the usurpation of power perpetrated by modeling and simulation, returning community attention to the issue by repeatedly pointing the way back to the deliberative process that once ruled over modeling and simulation is unlikely to be enough. Unless one proceeds to discredit actively the technique of modeling and simulation, its rule over deliberation will probably continue. A natural way of discrediting modeling and simulation is to point out its natural limitations and then show that questions fully relevant to the decision maker lie outside those limits. Revealing its inherent impotence to deal with anything beyond the mechanical task of replicating the war machine and the circumstances of its employment within the body of a computer, will unmask modeling and simulation for the fraud that it is, and hopefully will return it to the subservient position in which it belongs.
Discrediting modeling and simulation is no easy matter despite the self-evident nature of the task. The technique has the enthusiastic support of the analytic community because, unlike systems analysis, it requires little thinking; provided he includes in his simulation all factors, important or not, that he can think of, the analyst is no longer required to understand what is important to a given decision. In recent years it has also received the support of the military community because the non-deliberative character of modeling and simulation tends to assuage that community’s long-time distrust of civilian participation in their business by reducing their interaction with analysts to no more than the receipt of a recommendation. Finally, even the decision makers themselves appear to like the technique because its advertised completeness provides the same opportunity for passing the buck that the analytic community likes so much. Nevertheless, in view of its importance to the process of effectively controlling command technology, we must undertake the effort.
Before we proceed down this path, however, it would be helpful if we defined a little more precisely what it means to simulate a military engagement. The field of defense analysis is replete with examples of such simulations. They are very large computer models in which one tries to include all possible features of the engagement modeled. All military forces on both sides and the corresponding weapon systems are included as software objects, and their motions are generated by the model itself in accordance with a set of rules. When, from time to time, forces find themselves collocated, they interact with a probability that is determined by the geometry of the problem and is always the same in identical circumstances. The interaction is generally static in that the actions of one side engenders no free reaction from the other side; what happens in an interaction is determined by the creator of the simulation, not by the enemy. The outcome of each such interaction is the result of a model that the creator claims will represent the laws that govern the evolutions of such interactions, be they deterministic or stochastic.
And then the war goes on until, at some predetermined moment in time, the computer tallies-up the number of forces lost on either side and asserts the outcome of the engagement in terms of a number of possible measures that, while military sounding, are purely mathematical entities. These measures of effectiveness could be displayed for a variety of input parameter choices, but never for a variety of possible ways of modeling the interactions. In particular, alternative technologies, described as alternative sets of input values to the simulation, are explored by running the simulation with each alternative in turn and then comparing the outcomes. Absent any indication of what considerations were important and which of the assumption made about the world were crucial, this stark representation of technological choice can only indicate to the decision maker which technology to choose, not why he should choose that technology, rendering simulations unfit to participate in the conversation conducted around the decision-making table.
Since one can thus simulate only that which one can describe in this mechanical manner, modeling and simulation is forced to simplify the military processes that it models. It must ignore the possibility of a creative response from the enemy because creativity cannot be simulated. It must replace the dynamic evolution of a war with a static sequence of actions because, war being a free play between two deceiving entities, there are no general laws governing that dynamics for one to simulate. It must ignore the unique role that human judgment and flexibility play in the conduct of war because one cannot simulate a human being. It must provide answers to technology questions based only on the structure of a military force because it can not simulate any of the infrastructure activities involved in generating that structure, such as recruiting, training, and it can not account for the supporting structure designed to improve the life of military personnel and their families.
However, no matter what the war machine generated by the decision-making process turns out to be, it will have to be used by human beings engaged in a highly dynamic confrontation with a totally independent enemy, it will have to possess the flexibility to adapt to unforeseen circumstances to a degree generally unattainable by mere machines, and it will have to be supported by a fully responsive infrastructure of human activities designed to produce and enhance its effectiveness. Since modeling and simulation is used as a basis for technology decision, despite its inability to account for these considerations, one must approach such decision with a healthy measure of skepticism.
Incidentally, it is interesting to note, that the features of warfare missing from modeling and simulation correspond precisely to the destructive consequences of the Network Centric vision that we highlighted in Chapter 6. Network Centrism and modeling and simulation appear to go hand-in-hand, feeding off each other’s limitations. Therefore, by discrediting modeling and simulations we shall limit the general validity of Network Centrism claimed for it by the proponents.