effectiveness creates a dilemma in that neither the character of the programs nor the size of the budget can be determined initially. But that is a theoretical nicety, the direct consequences of which may not be of major importance. The vital point is the way in which budgetary limits may control force posture and therefore strategy. Shifting sands seems the best way to characterize the strategic rationales of recent years. In 1961 the suicidal implications of massive retaliation were underscored: the United States would be faced with a choice between humiliation or holocaust. Interest then developed in damage-limiting and coercion. But there has been little willingness to invest money in either. Since 1965 the merits of Assured Destruction have been emphasized-with little attention paid to the suicidal implications found so distressing in prior years. The principal rationale for the current emphasis on Assured Destruction reflects certain recently-developed notions of arms control. It clearly falls within the province of the decisionmakers to adopt a strategy of measured response to any Soviet buildup with the long-term objective of preserving U.S. Assured Destruction capabilities. One should note, however, that to accept this particular guide to action implies that the buildup of the Minuteman force in 1961-62 was a mistake. These newer arms control criteria may be the preferred ones, but they rest on the judgments and intuitions of the decisionmakers. They certainly do not emerge by themselves from analysis. May one infer that the oscillations in strategy have something to do with budget limits, or in this case something more specific: a preconception regarding how much this nation should spend on the strategic forces? I find the conclusion irresistible. The evidence antedates the current phase-down in the face of the Soviet buildup. Once again, these lie within the decisionmaker's prerogatives, but particular beliefs regarding budget limits or the "adequacy" of specific strategies should not be attributed to, much less blamed on, analysis. A USEFUL IF OVERSOLD TOOL Whatever resources are made available to defense (or any other mission), choices will have to be made. Allocative decisions inevitably are painful; many claimants will be sorely disappointed. Few will find fault with their own proposals, almost all with the machinery for selection. Any procedures for allocation will be criticized--even in a hypothetical case in which the conceptual basis is unarguable and no errors are made. Analysis provides the backup for a selective process. What does it contribute? How does it compare with real-world alternativesnot with mythical alternatives in which all claimants get their requests and no one is disappointed? It has been emphasized that analysis cannot determine the appropriate strategy. It can shed light on costs and tradeoffs. But the choice to press arms control or arms competition or to rely on tactical nuclears or nuclear firebreaks must be determined by the decisionmaker sustained primarily by hope, conviction, and prayer. Even if a decision could be demonstrated as correct at a given moment in time, there is the certainty that objectives will change over time. For these higher level problems analysis is an aid, but a limited aid. The toughest problems, dominated as they are by uncertainties and by differences in goals, do not yield to analysis. Happily many problems are more mundane and more tractable. Where analysis has proved its highest value is in uncovering cases of gross waste: points at which substantial expenditures may contribute little to any stated objective. It might be thought that a problem of diminishing returns exists for analysis in that the cases of gross misuse of resources are likely to be uncovered at an early stage. Thus, as the opportunity for major savings through elimination of irrational forms of waste theoretically recedes, analysis would be forced into the more ambiguous areas in which strategic choices become intimately involved. In some cases, where information is readily available and objectives and conditions relatively unchanging, this could prove to be true. The very success of analysis would then undermine near-term expectations of additional returns. However, in defense this turns out to be irrelevant, since the problems are so volatile and information so difficult to unearth. To say that analysis works best in cases of gross waste should not be taken to imply that analysis accomplishes little. The simple cases involving so-called dominant solutions may involve billions of dollars. The volume of government resources that may be lavished on the care and feeding of white elephants is simply staggering. Here we have "quantified common sense" in its most direct form. In bureaucracies, units at all levels are concerned with organizational health. Rather than making the hard choices, the tendency is strong to maintain morale by paying off all parties. Analysis provides a means for coping with this problem. The big issues may not be directly involved, though they are likely to be dragged in by the proponents of particular programs. Should the assessment of analysis be much influenced by the annoyance felt by those whose proposals have failed the tests? Certainly not in the general case. No more than should the decisionmakers be permitted to hide their judgments behind the camouflage of analysis, should the patrons of doubtful proposals be encouraged to argue that acceptance would and should have come-if only analysis had not been employed. Budgets are limited and hard choices must be made. If nobody were annoyed analysis would not be doing its job-of questioning both routinized practices and blue-sky propositions. Disappointment is unavoidable. The question is not the existence of annoyance, but to strive to annoy in the right way and for the right reasons. In this light it may be desirable to examine the issue of the generalist versus the specialist which has been touched upon in the Hearings. In the nature of things specialists become committed to particulars: a piece of hardware, a technological criterion, a disciplinary blind spot. It is a case of suboptimization run wild. Proponents of specific capabilities or gadgets tend to become monomaniacs. In a sense that is the way they should be: totally dedicated to their tasks. But one does not turn to them for detached judgments. There is no substitute for the informed generalist. There is a recognizable risk that the superficiality of the generalist may match the monomania of the specialist. However, that need not be the case. Although the generalist's knowledge cannot match that of the specialist in detail, analysis can once again play a useful role, by permitting the organization for the generalist of more specialized information than he alone could master. How does this relate to the limits of the analyst's role? Two distinctions should be kept in mind: that between the technical specialist and the analytical generalist and that between the analyst and the decisionmaker. The analyst's tools are not circumscribed by discipline or even by subject matter. But general tools are not immediately convertible into broad policies. Many analysts are, in some sense, specialists in the use of general tools. Being a good analytical generalist does not necessarily imply possession of such additional qualities as breadth, judgment, and political attunement. These latter qualities are what many have in mind when they speak of the generalist as policymaker. CONCLUSION In closing I should like to underscore three points. First, the position of the decisionmaker employing analysis is somewhat ambiguous. For tactical purposes this ambiguity may be deliberately augmented. Intermittently he may choose to stress analysis or judgment, and to shift hats according to the tactical_requirements of the moment. His policy judgments may be obscured or defended by cryptic references to detailed analyses which allegedly force the policy conclusions. On the other hand, if any limitations or inadequacies in the analyses should come to light, these can be waved away with the reminder that all issues are ultimately matters for the decisionmaker's judgment. Moreover, the pattern is in reality far more complicated than the standard exposition in which the analyst produces an objective study, and the decisionmaker's judgment enters at a later stage erected on the foundation of these objective results. That makes the analytical and judgmental stages seem clean-cut. Few studies are that pure. The decisionmaker's judgments quite typically are dumped in at an early stage in the form of guidance, assumptions, and terms of reference. The more political a study, the less likely is it to be pure. In fact, the process can be (and has been) far more corrupted, when questionable (phony) numbers are introduced. Since judgment and analysis are thoroughly intertwined in all but a few studies, the attempt of decisionmakers to shift roles by referring to fundamental analyses should be treated with some skepticism. The decisionmaker should not be permitted to escape the full burden of responsibility by the invocation of analysis. The temptation for those who have introduced analytical techniques into the government to treat their own positions or careers as identical with analysis is understandable. No outsider should yield to the same temptation. The roles and even the temperaments of decisionmaker and analyst are quite distinct. The confusion tends to disguise the heavy personal burden borne by the decisionmaker. More important, if analysis is treated as synonymous with particular decisions or personalities, there is a risk that it will be throttled or abandoned after their departure. From the standpoint of public policy this would be a major loss. Second, we should avoid the erroneous belief that the performance or potential power of analysis will be uniform in all contexts. If a town is considering building a bridge, a number of difficult analytical problems must be addressed: does demand warrant construction, where should the bridge be built, what should be its capacity, and so on. But once these questions are resolved the engineer falls back on a solid technical base. By contrast, for such goals as deterrence, assured destruction, controlled nuclear warfare, damage limiting, to say nothing of welfare benefits, we fall back, not on a firm technical base, but on what may be scientific mush. The distinction is not always appreciated. The difficulty is sometimes dealt with by referring euphemistically to the model problem. But our ability to formulate models depends upon our knowledge of the mechanics of the real world. For many problems our knowledge is meager, and the proffered models are misleading or downright erroneous. The lack of good models in many problem areas simultaneously limits the power of analysis, while increasing the burden placed on judgment. In treating analysis as a uniformly efficient problem-solving technique, the variability of analysis, which reflects the variability of the knowledge base, is ignored. Though analysis is a powerful tool, specific analyses vary greatly in quality. Some are little more than trash. But we need to discriminate, rather than to reject analysis in toto. At the present time there is some risk that we will do the latter. In an address some years ago Secretary Enthoven observed: "My general impression is that the art of systems analysis is in about the same stage now as medicine during the latter half of the 19th century; that is, it has just reached the point at which it can do more good than harm." That was a frank and realistic, if somewhat pessimistic, assessment of the state of the art. Scientifically speaking, there are numerous blind spots in medicine. Yet, most of us ultimately are inclined to accept the doctor's diagnosis, if not his advice. Quite plainly at the present time Congress and the public are having second thoughts regarding how much trust to put in systems analysis. No doubt it is necessary to develop a greater ability to discriminate. Nonetheless, I suggest that policy will benefit substantially from the analysts' diagnoses. Third, there is little doubt that analysis has been oversold. That strikes me as a rather standard result in matters political. But the reaction against the overselling could be more costly than the overselling itself. Analysis is a powerful instrument: with it our batting average has been far higher than without it. Analysis is also an adaptable instrument. The McNamara regime has in many respects been a highly personalized one. Its performance should not be taken as defining the limits of this flexible tool. Admittedly, analyses vary substantially in quality. Each should be taken with a large grain of salt. On the other hand, if one does not demand too much of it, analysis will prove to be a most serviceable instrument. O |