Back at MIT/Mailbag Follow-up

The Importance of Being Explicit

I just wanted to briefly note how great it is to be back at MIT- 23 years after starting my PhD there and not quite 16 since I finished it. While it may be true you can’t go home again- and Kendall Square has changed massively since I was last there regularly- MIT still gives me an immense warm fuzzy. One of the central reasons for that is the density of really smart people to talk to- many of whom disagree with me vociferously (one of those chaired my dissertation committee…) yet are open to spirited but not venomous argument.

I had lunch with one of them this past Thursday- Charlie Glaser. I have known Charlie for nearly 20 years and to put it mildly we disagree about counterforce and damage limitation. Yet we have co-edited a book together (along with Brian Radzinsky) and always have good conversations- he is super smart, highly logical, and very pleasant. At lunch Charlie noted he had just read my recent paean to damage limitation. He pointed out that I appeared to loosely conflate the concepts of “unacceptable damage” and “assured destruction.”

In his view assured destruction is an absolute measure- the damage required in a nuclear attack to cause the irreparable end of a state/society. More damage could be inflicted, perhaps, but it would be superfluous- making the rubble bounce, metaphorically. For Secretary of Defense McNamara and his analysts in the 1960s, 400 equivalent megatons delivered against the Soviet Union was assessed to be sufficient to achieve assured destruction. However, by the 1970s assessments of the speed and efficacy with which the Soviets might rebuild began to call such a criterion in to question. Regardless of the exact severity of damage required, the concept is independent of political stakes.

In contrast, “unacceptable damage” is inherently defined by political stakes. Almost any action that led to a single nuclear weapon detonating on a U.S. city would be seen as “a catastrophic blunder,” in McGeorge Bundy’s phrase. This means a single weapon could constitute “unacceptable damage” while being nowhere near “assured destruction.”

I found Charlie’s comments helpful and made the point that in circumstances where the United States would consider damage limiting attacks against a nuclear peer the stakes would have to be extraordinarily high. This would mean “unacceptable damage” might be quite close in magnitude- perhaps even equal to- “assured destruction.” In other words, damage limitation capabilities that reduced adversary striking power below the level of assured destruction would also reduce them below the level of unacceptable damage if the political stakes were viewed as high enough. This was a point I thought I had made clear by focusing my definition of damage limitation on adversary targeting criteria for what seems to be assured destruction (200 targets in the Soviet example I used) but Charlie is right that I was loose in my language. So to be clear- my definition of damage limitation is focused on adversary assessments of the requirements of assured destruction- but in the special case where the political stakes make the threshold for unacceptable damage high enough to begin to approach assured destruction. I had not made that condition as explicit as I should have.

In another case of not being explicit, when Pranay, Vipin, and I did the mail-bag episode of the podcast last week I did not mention two important points in responding to the listener who asked about “use or lose” pressures and damage limitation. Briefly, one important elements of my thinking is the concept of “optimum instability,” which was developed at the tail end of the Cold War by Glenn Kent (a relatively unsung brilliant thinker and doer in the Cold War) and David Thaler. As they note:

The most important conflict [arises] between the objectives of enhancing first-strike stability, on one hand, and extending deterrence and limiting damage on the other; i.e. the more robust the Soviets believe first-strike stability to be, the less they might hesitate to precipitate a deep crisis by engaging in serious aggression, for example, in Western Europe. Balancing between first-strike stability and extended deterrence presents a problem in the planning of strategic forces. . . . Indeed, one might argue that an optimal amount of first-strike instability is possible: that is, enough to deter the Soviets from generating a major crisis, say by invading Western Europe, but not enough to allow a major crisis to spiral out of control. Whether or not such an optimum actually exists, the concept provides the proper intellectual framework in which to think about the trade-off between first-strike stability and extended deterrence.

So the pressure that generates some degree of “use or lose” incentive is needed to avoid the stability-instability paradox. The listener was right to be concerned about those pressures- as Kent and Thaler note too much pressure might indeed eliminate space for negotiation.

However, the United States by the late Cold War was moving to a posture where much of its prompt counterforce capability did not reside in its land based fixed ICBMs. Instead it was embodied in the Trident D5 SLBM carrying the W88 warhead and deployed on Ohio-class submarines. This meant that the Soviets (or any other adversary) could not eliminate U.S. counterforce capabilities against any ICBMs the adversary withheld nor could it eliminate U.S. assured destruction capabilities.

W88 flight test on Trident D5 (image courtesy of DOE/NNSA)

This meant large scale adversary first use of ICBMs against U.S. ICBMs and bomber bases would consume much or all of its ICBM force. Even if this attack was successful, the United States could eliminate its remaining ICBMs with a limited number of Tridents while threatening assured destruction to ensure war termination. In contrast, adversary large scale first use of ICBMs against U.S. cities would all but ensure the adversary’s destruction from SLBM retaliation. Thus large scale adversary use would result in either its dynamic disarmament or would be suicide- not so much “use or lose” as “use and lose.” This helps to moderate the pressures the listeners was concerned about- and hopefully contributes to optimum instability.

I first begin thinking and writing about these points nearly two decades ago so it becomes easy in a podcast to gloss over them- it is just implicit in my thinking. But Charlie’s point made me reflect on the need to be explicit in concepts and thinking- so wanted to ensure our listener had a more explicit accounting of my thinking. That may or may not help.

Finally, on the glossary question- a long time friend in the U.S. intelligence community and now listener highlighted the glossary in the recent Defense Intelligence Agency publication Nuclear Challenges. I not only recommend the glossary but the entire publication. A comprehensive unclassified statement of DIA’s assessment of the nuclear challenges posed by Russia, China, North Korea, and Iran (though it predates this year’s events with respect to Iran)- well worth a read!