WAR GAMES The Secret World of the Creators, Players, and Policy Makers Rehearsing World War III Today By Thomas B. Allen McGraw-Hill. 402 pp. $19.95

The nuclear war game -- a simulated superpower conflict in which high-level players test war plans by means of computer analyses -- is a part of the lore of the atomic age. Along with Dr. Strangelove and the "black bag" of war codes that accompanies the president, war games are the subject of fascination among the educated public who concern themselves with the problems of national security and danger. In my own experience students seem to have a morbid curiosity about what happens in these games. What do we do if the Soviets invade China? What happens if a terrorist sets a nuclear bomb in Washington when a crisis is going on somewhere else?

Public interest in the subject is different from that of the insider professionals who actually design these games, if Thomas B. Allen's book "War Games" is a fair description of the war-gaming industry today, as I believe it is. Inside this secret world, process dominates subject matter, and there is greater interest in perfecting arcane methods than in nailing down conclusions about preferred policies. More fundamentally, there is a kind of relaxed everydayness attached to the business, and a view that it is indeed a business with marketing fads, competitive infighting and nasty comments about the field as a whole. For insiders, war gaming develops a familiarity and routine that remind one more of managing a video store or cheese shop than of running a nuclear crisis.

Allen's book is an interesting account of war gaming today, three decades after the excitement of the pioneering work in the 1950s has worn off. Back then the problems were new, because nuclear weapons were new. People apparently thought nuclear war was something of a possibility. No longer, as Allen lets us know. Current sophisticated thinking discounts the likelihood of the big war, and one of the problems in war gaming today is to get the game to go nuclear. Allen asserts that Pentagon games have to be rigged to get a nuclear escalation, and that one advantage of replacing human beings with role-playing computer programs is that escalation becomes much easier. My own experience in several gaming exercises supports Allen's observations.

The strength of "War Games" lies in its insights, like the difficulty of getting human players on either the Soviet or American sides to initiate nuclear war. Games capture insights and experiences that are not written down anywhere else, so books such as this actually become important vehicles for communicating within the large defense establishment. In a European war the Soviets detonate a high-altitude nuclear explosion, the radiation from which knocks out NATO's radios and radar. NATO then must negotiate in the blind, or escalate without knowing what's really going on. Or consider the Soviet nerve gas attack in a different war also played in Europe: "Very few of the base personnel had managed to get into their protective gear in time ..." Both the command-and-control blackout problem and the inability of NATO forces to protect themselves from chemical attack are real concerns that might be ignored even more than they are were it not for the forced attention arising from their examination in war games.

But internal communication for the purpose of standardizing threats and strategies is not the only useful domain of war gaming. Allen points out that in certain cases games are played and results intentionally leaked for Soviet consumption.

A 1982 game, code named Ivy League, involved a crisis in northeast Asia that escalated to the nuclear level in the Atlantic. A Soviet missile attack on Washington followed, aimed at killing the president and thereby paralyzing the U.S. retaliatory blow. However, in this game the player acting the role of the president ordered the vice presidential player to go airborne aboard the so-called doomsday aircraft that serves as a command post for nuclear war. The vice president, safely airborne, then ordered a massive counterattack against the Soviet Union. The point here is that gaming can be used to test internal procedures, but it can also be used to bolster deterrence against Soviet actions. Regardless of what the Russian military may think, they now must seriously understand that paralyzing nuclear attacks aimed at the United States may backfire in spectacular fashion.

"War Games" is full of anecdotes, comments, digressions and after-hours gaming stories that are the best part of the book. These after-hours tales often contain much more interesting material than do the official game reports written within the bureaucracy. They give an intuitive feel for how people face what at bottom are impossible decisions, and they are not restricted to committee review as are official reports. Official committee reviews often have to remove all material conclusions that might be offensive, and they are dominated by consideration of process and method rather than subject.

Allen also does a good job of politely describing an all-too-common gaming phenomenon of the 1980s: the rigged game. Some games are designed by tamed consultants to endorse pet Army or Navy views, the reward for which is often a follow-on consulting contract. Indeed, within the gaming industry there are some large annual simulations whose outcomes are so tightly controlled by their sponsors that they are widely viewed as advertising campaigns.

Thomas Allen's "War Games" may strike some as a bit disorganized, without an overarching theme connecting the scores of games and research centers that he describes. However, this is actually a feature of the gaming field in the 1980s, and is something the author cannot be faulted for. His book should be seen as an interesting update on a fascinating and important subject. The reviewer, professor of public policy at Yale University, is the author of "The Command and Control of Nuclear Forces."