Secretary of Defense Caspar Weinberger and his deputy secretary, Frank Carlucci, are gradually developing their approach to managing the Pentagon. Their theme is decentralization, leaving it up to the Army, Navy and Air Force to manage their respective organizations and the resources under their control, while top civilian officials concentrate on "overall strategy."

This plan is completely inconsistent with three decades of experience trying to control the military-industrial complex and its claims on national resources. Indeed, the prospect of unprecedented growth in peacetime defense outlays, combined with mounting evidence that the costs of weapons are dangerously out of control (with much of the expensive "gold-plating" occurring after the secretary has given the service the go-ahead), seems to call for more, not less, active management by the office of the secretary of defense.

Controlling weapons costs is hardly a new problem. Studies conducted over two decades ago documented the 200 percent to 300 percent cost overruns that resulted when civilian leaders left it to the services to manage weapons acquisition. Defense secretaries from Robert McNamara through Harold Brown have sought in various ways to keep weapons costs under central managerial control. McNamara tried total package procurement. Melvin Laird and his deputy, David Packard, initiated a "fly-before-you-buy" policy. The results have been mixed, at best, the most apparent consequences being, unfortunately, complexity, confusion and delay in decision-making. Management and managers in general have a badly tarnished reputation in DOD.

Thus the most perceptive critics of the department are citing the attempts to manage the weapons acquisition process as a primary cause of recent problems. James Fallows, for example, in his otherwise valuable book, "National Defense," argues that "managerial minds" and "managerial logic" are the root of the problem of unnecessarily complex, poorly designed and excessively costly weapons. Such criticism finds a receptive audience among those who believe that DOD is just another General Motors and that the services can produce aircraft, ships and missiles just like GM's decentralized divisions produce Chevrolets, Pontiacs and Oldsmobiles.

This view misses the whole point of what we have learned from recent experience. The process of conceiving ideas for weapons, translating the ideas into "requirements" and moving them through various stages of development into procurement and production is the most decentralized, indeed subterranean, of government activities in the province of the services and their contractors. The whole structure is protected by layers of military politics and by members of Congress with military bases and contractors in their districts. The weapons acquisition field is virtually management-proof. Decentralizing this aspect of defense management, then, will be one of the easiest things Weinberger and Carlucci undertake. All they need do is sit back and watch.

They will not like the results, though. The weapons proposals now in the pipeline are likely to yield the same expensive, poorly performing products as those Fallows so ably documents in his book--weapons that reflect an unreasoning obsession with maintaining technological superiority over the Soviet Union, that are so complicated that they break down constantly, that are expensive nightmares to maintain, and the designs of which are based on ludicrous assumptions about how soldiers actually behave in combat.

Such results are likely to create two serious problems for the Reagan administration. First, congressional committees are bound to insert themselves ever more deeply into defense decision-making to ensure that elected leaders somewhere are looking out for the taxpayers. Second, the almost certain continuation of the procurement fiascoes, which have become the trademark of Pentagon management, will become an albatross to an administration that promised sound, tough-minded management.

Weinberger and Carlucci may feel compelled to preach the rhetoric of decentralized, business-like management, but public managers as able and experienced as they should avoid its seductions. As initial steps, they should consider the following:

First, they should carefully but expeditiously study the record of weapon systems acquisition management to discover what has worked, what hasn't and why. People who know how to analyze this kind of experience and produce useful lessons for managers should be asked to do so--soon.

Second, Weinberger and Carlucci should mobilize the defense analysis community inside and outside the Pentagon to provide them with ideas, constructive criticism and good analysis of alternatives on a continuing basis.

The key to effective leadership in improving America's military posture is access to a wide range of thinking and a willingness to pursue ideas, approaches and philosophies other than those that survive intraservice political struggles. They must create that access for themselves before the opportunity vanishes.

Third, in actively thinking through the issues and alternatives concerning weapons design and acquisition, they should work assiduously to build a bipartisan coalition of interested and knowledgeable legislators conversant with the problems they face. If relationships with Congress are limited to confrontations over proposals on which the Pentagon is already dug in, the scene of creative thinking will shift to Congress. The likely result is continued confusion in military-force planning.