If, after a nuclear war, there were anyone left to write its history, the final blow to sanity would be learning that it was all an accident, that some misunderstanding or stupid error had wiped out civilization. Almost as bad would be the discovery that bravado or bluff rather than a real attack had led to Armageddon.

William L. Ury, coauthor of the negotiator's handbook "Getting to Yes," has proposed a possible way to avoid at least the blooper path to annihilation. A permanently staffed, 24-hour joint crisis control center, he argues, would allow trained Soviets and Americans who know one another -- and that is key -- to talk over developments face to face before either side pushed the button.

But Ury confuses techniques of negotiation with the mechanism of arranging talks. His book makes a useful suggestion about setting up a permanent talkfest, but muddles the discussion of how it would work with advice on diplomatic negotiating tactics that both sides should be taught in case a crisis arises. He really wanted to write the book on how to avoid nuclear war altogether, but this isn't it.

Ury says the Washington-Moscow hot line -- really a glorified teletype -- is good as far as it goes, and that its recent upgrading to allow instant exchange of graphs, maps, charts and so on was a major step forward. But crises before and after the hot line was installed suggest scenarios it might not be able to handle. For example, the hot line did not manage to keep the Korean Air Lines 007 jet from being shot down in 1983 when it flew over Soviet territory.

Ury notes that the top U.S. and Soviet military leaders have not met since the end of World War II, except briefly at the SALT II signing. Without any face-to-face experience, they cannot judge one another's intentions except in the remotest and most theoretical way. Mutual suspicion reigns worldwide.

This became evident in the Cuban missile crisis of October 1962, when the Russian and U.S. leaders could not communicate fast enough through existing channels. They had to use an ABC news correspondent, midnight restaurant meetings and open radio broadcasts to deliver crucial messages at crucial moments. Their experience led to the hot line, which was employed during other face-offs in 1967 and, especially, in the Mideast war of 1973.

That was the real scare. The superpowers "partially lost control of the situation to their respective allies, who were far more concerned about their immediate conflict than about the risk of U.S.-Soviet escalation." Ury observes, rightly, that another such conflict among our respective clients is likely, and that any one of them could drag the superpowers into blowing each other away.

The implication, especially with nuclear weapons proliferating around the world, is that the Soviet Union and the United States could arrange things through their permanent talks so that they will both just stand back and watch the minipowers blast each other. That would undoubtedly be preferable to world nuclear holocaust, but as Ury notes, consultations "could easily be misperceived as superpower collusion. Washington and Moscow would need to inform and consult their allies expeditiously, especially about discussions directly affecting their allies' interests."

The KAL jet disaster, which killed 269 presumably innocent people, is Ury's prime horror story of a situation that a crisis control center might avoid by exchanging evidence that no espionage or aggression was intended. But there is much suspicion, even on the U.S. side, that in fact the plane was on a spy mission. The only way a crisis control center could operate in this case, if the plane were spying, is if the negotiators did not know the truth. And once that came out, the center would lose its credibility.

For cases where there is real aggression, Ury suggests that leaders adopt a "stabilization" strategy of the kind that came up with the Berlin airlift in 1948, essentially an end run around the confrontation at the Berlin border. The 1962 naval blockade of Cuba was another end run that avoided confronting the Soviet missiles directly but allowed Russia to save face.

It is hard to see how a crisis control center could help construct end runs if its people knew all the inconvenient facts. Clearly there are situations where it would be ineffective. In fact, there would be situations where peace could be maintained only by publicly insisting that a deliberate act had been an accident.

In spite of Ury's fog on these points, his suggestion for a place for permanent U.S.-Soviet talks at the secondary level has a lot of merit. If we're going to go out with a bang, let it at least be over something serious.