In our thinking about preventing a nuclear war between the United States and the Soviet Union, more emphasis should be given to arms control measures that contribute to crisis stability. Actions that allow each side to maintain coherent political control over their own nuclear forces, meas enable each to understand the significance of dangerous operations, and systems that help to terminate a breakdown of deterrence all need more attention than they have received to date. Steps have been taken in this direction, but more can be done on this underside of arms control.

Significant advances in arms control have gone largely unnoticed in this regard. Attention may center on summit-level arms agreements that constrain force levels of missiles and warheads. Yet the initiatives taken by the U.S. Department of Defense and apparently the Soviet Ministry of Defense to drive the probability of accidental nuclear war down to zero is one of the paramount arms control achievements of the nuclear age.

Today, the chance that a flock of Canada geese crossing a radar screen or a failed computer chip could start a nuclear war is effectively nonexistent. Scare stories to the contrary, in peacetime there is no danger of an accidental nuclear war, a conclusion reached by all serious investigators of the problem. The current situation was not automatic, for in the early days of nuclear weapons the danger of accidental war was considerable. Yet through large investments in hardware and careful organization of our nuclear forces we have managed this problem well.

The problem we now face can be described as one of contingent danger. For the success of managing the problem of accidental nuclear war in peacetime contrasts with the difficulty of doing so in a crisis. The differences are considerable, for in a crisis our expectation of attack increases, and many of the peacetime controls that prevent accidental war are removed when strategic and theater forces are placed on alert. That, after all, is what it means to go on alert. In addition, in a political crisis the danger of inadvertent war arises, a war resulting from an escalation process in which each side keeps escalating over the other until an "eruption" occurs that was not itself intended.

It must be admitted that as long as things remain peaceful, reliance on the current state of affairs will work well enough. That is, if the United States and the Soviet Union can avoid direct confrontations, then the initiating trigger needed to precipitate an accidental or inadvertent war in a crisis does not occur, and war will be avoided because conflict by accident in peacetime has been eliminated as a serious problem.

Two policy approaches follow from this observation. First, it may be best to concentrate our energy on preventing confrontations, by diplomacy, wise foreign policy, and the fostering of a cooperative relationship between the United States and the Soviet Union. This was the rationale for the policy of d,etente followed by the United States in the 1970s. In the future we might work to overcome near-term obstacles to use START-type negotiations to build a framework of cooperation between the two nations.

The second policy approach is to reach into the structure of the problem itself. That is, to establish rules of the road so that if an intense crisis involving high nuclear alerts does occur or if deterrence actually does fail, an eruption to large nuclear war is rendered less likely because of accident or inadvertence. This will not be easy, because it requires striking a balance between the twin dangers of accidents, which lead to a loss of control, and overly cautious actions, which could compound our security problem. But the obstacles may not be any greater in the long run than those involved in restoring a cooperative relationship between the United States and the Soviet Union.

These two policy approaches should not be seen as competing alternatives. Talk of "setting priorities" or establishing an agenda of importance presumes that there is an underlying standard of comparison between alternatives, whereas in this case the two approaches are working at very different levels of the problem of preventing nuclear war.

Some may interpret any thinking given to crisis actions in high alerts or even in a nuclear exchange as an indication of a warfighting, rather than a deterrent, doctrine. Yet it is this attitude that has left the problem of inadvertent war to narrow military staffs without broader political review of alerting operations, dangerous strategies and foolish military assumptions. An exclusive focus on deterrence may reinforce the danger of inadvertent war in crisis because it encourages unrealistic and even sloppy planning.

There are broader reasons still for tackling the accidental and inadvertent war-in- crisis problem directly. We should attempt to build security institutions whose success in avoiding nuclear war does not depend on crisis avoidance and a cooperative political spirit. What is needed is a system capable of withstanding adverse political relations, and variations in the political stripe and the competence of leadership -- and perhaps even gross stupidities.

We have designed our deterrent against direct nuclear attack on this basis and should put our actem on as firm a footing. However difficult the problems, we should be thinking about long- term nuclear policies, anticipating that we are not likely to achieve general nuclear disarmament even by the early 21 century.

A first step in this process is to recognize that the problem of accidental or inadvertent war will be difficult to resolve if we attribute it all to incompetent people. Leaders have made mistakes in past crises and the military has not always carried out its orders suitably. But it is unproductive to speak of stupidity, the use of invalid mental models, or the introduction of "biases" in crisis decision-making as if we somehow understood how to overcome these obstacles. Our understanding of human behavior and its modification is far too shallow to accomplish this worthy objective.

Instead of trying to change people, it will be more effective to change the premises of their decisions through better organizational design, information flows and removal of the threats that compel them to make irrevocable choices without due consideration of alternatives.

For example, trading deployments of Pershing II missiles in Europe for the Soviet nuclear submarines near the American East Coast would give both sides precious minutes to take such steps as searching for corroborating evidence of attack or even translating messages sent over the Hot Line.

Similarly, Robert McNamara has proposed that second use of nuclear weapons not be authorized until it is absolutely certain that an attack has taken place. The deployment of reliable satellites and ground-based warning systems can make this possible, and progress is being made on this front.

Other ways to improve our ability to deal with crises can be imagined. Realistic simulations that introduce members of the political high command to the problems of intense crisis and even the breakdown of deterrence are needed. Our political exercises should introduce the fog of war, the breakdowns of command that occur in the real world, and the complexities of coalition defense. At the White House level there is a real need for continuity of people who understand the alerting process and its possible consequences.

Mutual Soviet-American agreements could reinforce confidence in crisis stability. For example, each nation could pledge noninterference with the national warning sensors of the other, paralleling pledges of noninterference with national means of verifying arms control agreements. Of course, we could not guarantee that this pledge would be honored. But we could guarantee that the Soviets would know just how dangerous such interference would be in a crisis.

The list of suggested improvements and questions could go on, and the answers may be unpleasant or infeasible for various reasons. The goal of this exercise, however, is not to construct bizarre scenarios or even a list of arms control measures for its own sake. The purpose is to reduce the increased probability of accidental and inadvertent war in a crisis.

Unless a more sober attitude toward this problem emerges, the world may be edging toward an institutionalized nuclear showdown, one that admittedly is contingent on an external political shock to set it in motion, but one that may demonstrate how irrelevant many of our strategic and arms control ideas are to the security needs of the late 20th century.

The writer is an associate professor in the School of Organization and Management at Yale University.