When a man has knocked his head against the wall several times and still has failed to get through, it is perhaps time to stand back and think again.

The MX missile program was first given the go-ahead in 1974. Since then, it has swallowed dollars, human ingenuity and the political energy of at least two administrations, in the so-far vain attempt to come up with a concept of how to deploy the missiles in such a way that they would be largely immune to Soviet attack. The strategic analytical community provided the intellectual justification for these efforts by inventing that intriguing, persuasive term, "the window of vulnerability."

One can understand why it has been so difficult for political leaders to resist the lure of this slogan. For it implies not only that the vulnerability of U.S. missiles to Soviet attack is strategically intolerable, but also that there is hope that the window can be closed, that vulnerability is, at worst, a temporary problem.

Both these implications have, however, become increasingly questionable. There is no doubt that it would be highly desirable if U.S. strategic missiles on land could not be destroyed by a preemptive enemy attack. But do strategic deterrence and stability really disappear if only part of the U.S. strategic arsenal, carrying less than 25 percent of the total of nuclear warheads, could be destroyed by a Soviet strike? Those who maintain this view have to argue that to attack U.S. ground missiles alone would confer on the Soviet Union a concrete, sizable strategic advantage; it is more likely, however, to entail such enormous risks for the Soviets -- in execution and in consequence -- as to be meaningless once the threat of nuclear war has concentrated minds.

The second assumption behind the "window of vulnerability" is even more questionable: that somehow there are ways to restore invulnerability to fixed land-based missile forces. The president's proposal for the "Dense Pack" basing is only the latest and probably not the last in a series of schemes as ambitious as they are uncertain and unaffordable. Perhaps tomorrow there will be "Air Pack" -- the MX on giant aircraft -- or even "Space Pack," and the more each of these proposals reveals its shortcomings, the more it seems to spur the search for even more complicated technological fixes. So much is clear by now: if there really is a window of vulnerability, its shutters have very rusty hinges, and they have proved, to date, extremely resistant to being closed.

This is probably not for any lack of technological ingenuity. More likely, the vulnerability of all military installations that cannot be moved about to evade detection has come to stay. Indeed, the image that the supporters of the MX program have created may be more apt than they thought: the window, rather than being closed, may offer a view to the future conditions of strategic deterrence, the end of the age of invulnerability and the dawning of that of vulnerability.

To date, the United States relies for over 75 percent of its strategic capacity on weapons systems that are still relatively immune to attack -- the submarines and the bomber planes. They are likely to remain so for some time, but not indefinitely. There is time, therefore, to contemplate what it will mean for strategic deterrence if the weapons of last resort that serve it are no longer immune to destruction, what new requirements this will raise for defining stability, and what new tasks it will pose for arms procurement and arms control.

The dogged pursuit of ways to provide protection for fixed weapons on the ground is not only likely to fail, it also leaves us less time to tackle this more important, conceptual task. As military technology moves increasingly ahead in making available the means of detection and precise delivery, the vulnerability of all military targets will become the rule, invulnerability the exception.

Analysts and politicians alike should use the time to ask if they have not tended to over-sophisticate what is, after all, the rather primitive notion of nuclear deterrence: that an enemy will be prevented from attack by the credible threat of devastating nuclear retaliation. Practitioners and experts alike have perhaps fallen victim to the temptation of over-refining the essentially unrefinable.

It is said that each leg of the "Triad" -- submarines, bomber planes and land-based missiles -- would have to be invulnerable; but why could deterrence not rest adequately on two legs and a half (as it does in the Soviet Union)? Why should it be necessary to be able to respond to every strategic option that the Soviets have, if discrete nuclear options are a contradiction in terms? Why not undercut Soviet programs aimed at hitting fixed U.S. targets by reducing the number of those targets?

The problems raised by the MX are a forceful reminder that, rather than blame the technologists for their inability to come up with convincing remedies to our conceptual problems, we have to consider the concepts themselves. If that lesson is learned from the present MX predicament, the millions of dollars so far invested in the program will not have been entirely wasted.