THE ONE THING almost anyone will say about Afghanistan is that the United States cannot just sit on its hands. But why not? If Washington cannot prudently do something that is directly helpful -- for example, free Afghanistan -- then it should be making the least of it, not the most.

And if the United States cannot defend the Persian Gulf from halfway around the world, if all it can really do is threaten to blow up the planet, then it should be hedging against the deprivation of oil, not planning to start a world war.

The logic of nonintervention is ill understood, even by those who are assailed for their pusillanimity. The trouble is not that they are too isolationist but that they are not isolationist enough.

In a world interlaced with risks of nucler destruction, the United States should be looking for ways to minimize provocations, to let regional conflicts burn themselves out. With respect to American allies and clients, the prescrition would be disengagement for the United States, self-reliance for them. Foreign leaders seem to understand this much better than Americans.

An example is Pakistani President Mohammed Zia ul-Haq, Zia knows that those who have depended on the United States must now make up deficiencies in their own defense posture, and perhaps simultaneously make some intelligent accommodation with powerful neighbors. On Jan. 14, 1980, he told Newsweek: "Given the new power equation in the world, self-reliance will be the key to our survival as a nation . . . If you have to live in the sea, says an old proverb, you have to learn to live with the whales." Accordingly, Zia spurned the belated military aid package offered by the United States in the wake of the invasion of Afghanistan.

The trick is to learn to live in a world without deals, without rules -- an asymmetric world, an unfair world. Bleak as it may sound, Americans should observe a code of conduct that is constructive for themselves, even if it is not reciprocated.

The key is acceptance of the costs of nonintervention. Self-described hardliners, rarely resort to hard data about the costs of the moves they urge upon the American people. Instead, they talk about self-hatred and the culture of appeasement. The debate should be about the requisites of peace in a nuclear age.

A policy of strict nonintervention recognizes that the United States might have to let some countries be dominated by its adversaries or their supposed proxies. A guideline for nonintervention is a security perimeter based on geographical and functional considerations.

The georgraphical considerations are the most comprehensible. For example, neither the Persian Gulf nor the Middle East in general would be within America's sphere of active defense. This is not to deny that the United States has interests in this area, but it would not defend those interests by force or military threat. Waging or even risking nuclear war to reserve American access to countries and their resources is not an appropriate foreign policy for the last two decades of the 20th century.

Western Europe is obviously an object of critical American interest, but it would also be excluded from America's security perimeter. Ironically, because the cases America cares about most are also most demanding and possibly terminal, they are the ones the United States would in fact -- and should in principle -- shrink from defending.

The United States would defend against direct threats to those values that are so basic that they are part of the definition of state and society: the autonomy of its political processes and the safety of U.S. citizens and their domestic property. It would also defend against challenges that are indirctly threatening but are relevant because of their weight, momentum, direction and ineluctability.

This may not be a very sympathetic position to hold. But if anything must change -- if anything has in fact been changing -- it is the ability of the United States to do something effective about the rest of the world and the presumption that it ought to do something to prevent one kind of outcome and promote another.

The largest question is whether time and circumstance have outrun the 30-year American paradigm of deterrence and alliance and whether a new orientation is required. The United States must adjust policies to events, not events to policies.

It must ask not whether intervention is justified -- of course it sometimes is -- but whether intervention is necessary, whether the object of the intervention is part of the irreducible core of U.S. security. If it is, America has no choice but to defend. But if the United States can live without something -- and most situations are in that category -- then it must be disposed not to become involved, especially since chalenges will not always occur in convenient forms and places.

Oridnary Americans are constantly evaluating whether the game is worth the candle -- not whether the game is worth playing at all. The underlying question, therefore, is whether American leaders can count on their own people to generate this candlepower, not in placid and untested moments but in critical times.

That is why accusations of failure of nerve and invocations of will are so shallow and misleading. Will is not something foreign policy players can summon and shape and project, but is the complex of domestic constraints, which are as foreign and inaccessible as the recalcitrant behavior of other countries.

What are the costs of intervention and nonintervention? There are some important intangibles. If the United States insists on peace to the extent of appearing to default on its commitments, America might invite the defection of some clients from its security protection, the dissolution of alliances, the loss of American control and a looser international system.

However, the preparation for war is costly, in terms of public and private welfare diminished, constitutional processes distorted, citizens and their assets mobilized, bodies and minds regimented. These costs are borne year in and year out, whether the eventual intervention is a quagmire or a piece of cake.

What are the more tangible costs? Out of the fiscal year 1981 defense budget of $159 billion, the United States is already spending some $25-$30 billion to support its interests in the Middle East, the Persian Gulf, Southwest Asia, and the Indian Ocean. How much more would it cost to meet the demands of the hawks -- and the requirements of the Carter Doctrine -- to provide more strenuous defense and deterrrence in these new areas of American commitment? And what would be the costs of failing to do so?

Actually, estimating the costs of war and peace in this area is an arbitrary exercise. Preparation for war might cost an additional 8 percent of the defense budget compounded for 20 years, or $587 billion. A Vietnam-size regional war would cost about $450 billion. There is also the additional chance that it might escalate to general nuclear war, in which the United States might lose half its annual gross national product, or $1.25 trillion, for 20 years.

All this must be thrown into the balance against the consequences if the United States failed to defend and if some sequence of events resulted in the deprivation of oil. America might lose 20 percent of its GNP for 20 years.

In the last analysis, the costs of defending or abstaining are incommensurate. One way, the United States edges perceptibly closer to the devastation of war. The casualties of a general nuclear war might be 125 million Americans; the casualties of even a conventional regional war might be 50,000 to 100,000, and it would lose the oil anyway. The other way, Americans would live less well, but live.

In the aftermath of Afghanistan, the comfortable middle options have dropped away. The only alternative now to the official strategy of resuscitated military interventionism is an isolationist foreign policy.