War Games and Nuclear Risk

One way to apply risk analysis to a potential failure of nuclear deterrence is to decompose a catastrophic failure into four steps and estimate the risk of each step: 

• the rate of occurrence of initiating events, which have the potential to produce a crisis between two nuclear powers,
• the risk that the initiating event produces a crisis,
• the risk that the crisis crosses the nuclear threshold, and
• the risk that, once the nuclear threshold is crossed, the conflict escalates to all-out nuclear war.

Historical data exists for estimating the first two parameters, but estimating the last two requires subjective evaluations. In my preliminary risk analysis, I used a crude form of expert elicitation. I would have liked to incorporate data from war games, but was unable to find any at that point in time. Several sources have since become available, and indicate that those risks may be dangerously high.

A 2008 RAND Project Air Force report states:

In 2004, Director of Air Force Strategic Planning Major General Ronald J. Bath sponsored a war game in which uncontrolled escalation occurred, surprising players and controllers alike … this experience was just one in a series of escalatory events occurring in major war games over the past several years.

A 2012 report by the lead author of the above report, RAND’s Dr. Forrest E. Morgan, added a few more details:

By 2004, strategic planners at Headquarters U.S. Air Force had become concerned that they did not adequately understand escalation risks in the contemporary security environment. … An increasing number of war games … had ended in uncontrolled escalation, games in which the scenarios called for only limited U.S. military intervention against notional adversaries that were clearly outmatched by U.S. forces. … At first game analysts assumed the outcomes were spurious, the result of overly aggressive “red teams” … But the increasing frequency with which the games turned escalatory and the wide range of participants and scenarios suggested that something else was at work, something that Air Force planners did not understand. 

While classification prevented more details from being disseminated, Yale Professor Paul Bracken’s excellent, just-released book, The Second Nuclear Age, reveals significant details of a June 1983 war game, codenamed Proud Prophet (pp. 81-89).* This war game differed from earlier exercises in that Secretary of Defense Caspar Weinberger and JCS Chairman. General John W. Vessey Jr. played themselves and our actual war-fighting plans were used. According to Bracken:

The result was a catastrophe that made all the wars of the past five hundred years pale in comparison. A half billion human beings were killed in the initial exchanges and at least that many more would have died from radiation and starvation. NATO was gone. So was a good part of Europe, the United States, and the Soviet Union. Major parts of the northern hemisphere would be uninhabitable for decades. (page 88) … 

This game went nuclear big time, not because Secretary Weinberger and the chairman of the Joint Chiefs were crazy but because they faithfully implemented the prevailing U.S. strategy. (page 88) … 

after Proud Prophet, there was no more over-the-top nuclear rhetoric coming out of the United States. Launch on warning, horizontal escalation, early use of nuclear weapons, tit-for-tat nuclear exchanges – these were banished, conceptually and rhetorically. The Reagan administration switched gears. The chairman of the Joint Chiefs spent the next several years cleaning up U.S. war plans. Nuclear threats were gone. (page 89)

But is the nuclear threat gone today? The 2004 war games, described above, seem to indicate that threat is alive and well, as do recent implied US nuclear threats against both Russia and China – see my blog posts of November 10, September 28, and September 26 for details. And, as Prof. Bracken’s book ably demonstrates, nuclear proliferation and terrorism have added dangerous new dimensions.

It should come as no surprise that, when deterrence fails, it has a tendency to blow up rapidly and unexpectedly. The very name “nuclear deterrence” assumes that nuclear threats will deter. But nuclear deterrence can morph rapidly and unexpectedly into the related game of “nuclear chicken,” in which each side resists behaving rationally for fear of its “nuclear deterrent” losing all credibility in future crises. We need to avoid not only crises, but the potential initiating events which can lead to them. The earlier in the accident chain we catch the process, the safer we will be. My briefing paper on the 50^th anniversary of the Cuban Missile Crisis lists eleven such accidents waiting to happen, and proposes ways to reduce the risk of their leading to crises.

Martin Hellman

To subscribe to this blog, enter our URL (https://nuclearrisk.wordpress.com/) into your RSS reader.

To learn more about the level of nuclear risk, visit my related web site.

I am grateful to Dr. Tony Barrett of RAND for alerting me to the RAND reports referred to above.

* An end note on page 280 of Bracken’s book explains why Proud Prophet has never been described earlier:

The Proud Prophet war game has not been disclosed before. Phillip Karber promised Secretary Weinberger that he would not discuss any aspects of the simulation for twenty-five years, that is, until 2008. The account here draws on my recent discussions with Karber and my notebook of observations recorded at the time in 1983. I am very grateful to Phillip Karber for his permission and cooperation to describe this game.