In light of modern authoritarian regimes armed with nuclear weapons, it is important to reexamine what variables may “keep a limited nuclear war limited”.1 Traditionally, nuclear game theory scenarios revolve around the security of the state and any corresponding nationalist motivations, but this model may be incomplete. The primary motivation of any U.S. adversary in a potentially nuclear conflict may not be the security of the state or even the contested objective, but the security of the regime itself (or party control).2 The current international security situation is tense: given recent behaviors, the risk of nuclear use has arguably grown to a level not seen since the Cold War. If this “party control” assumption is true, it is important to consider the ramifications in a limited nuclear context.3

Nuclear war is inherently irrational in the traditional strategic context: “winning” even a limited war may involve massive losses. No offensive nuclear use can occur without considering what the response of the adversary will be, regardless of who uses nuclear weapons first. Using the “party control” assumption, three variables might determine whether an U.S. adversary, after absorbing a nuclear detonation, elects to de-escalate, retaliate, or escalate. These options are defined as such:

  1. De-escalate: the adversary does not respond with a nuclear attack. The objective may still be contested, but only with conventional means.4
  2. Retaliate: The adversary responds with a “tit-for-tat” nuclear strike or significant conventional effort.
  3. Escalate: The adversary responds with an attack that inflicts greater harm than the U.S. strike and increases the likelihood of more nuclear use.5


The three major variables that might determine how nuclear weapons will be used in a limited nuclear war are target suitability, the authoritarian’s sense of stability, and messaging. The first two variables occur on parallel tracks but are not necessarily related:

  1. Target Suitability: If a tactical objective cannot be met with a nuclear weapon, then “de-escalation” (at least with regard to nuclear weapons) is the rational choice. For example, in a vacuum, a rational actor will not respond to a nuclear strike with nuclear weapons if there is no tactical objective that can be achieved. Such a strike would accomplish nothing and would invite further nuclear attacks.
  2. Authoritarian’s Sense of Stability: the most important variable is the strategic determination of whether the adversary’s grip on power is threatened. If, despite the American use of a nuclear weapon, the adversary does not believe that the U.S. is attempting regime change, then either de-escalation or (if a tactical benefit exists per the first variable) retaliation is the most rational response because it does not necessarily invoke a second U.S. strike. However, if the U.S. nuclear strike somehow undermines the adversary (by emboldening internal opposition, for example), the authoritarian’s grip on power has been threatened. Retaliation is now the minimum rational response, and the risk of total nuclear war has increased.
  3. Certainty of Information and the Transparency of Motives: regimes must be confident that their legitimacy is not being threatened for retaliation to be the maximum rational response. If the United States introduces uncertainty by taking actions that could be inferred as threatening the regime directly (such as destroying additional military bases or launching nuclear capable bombers), then the adversary’s maximum rational response is now escalation, and the risk of total nuclear war has increased.

These variables are important because, if for whatever reason, the U.S. is compelled to use a nuclear weapon, the war will only “remain limited” if the adversary is inclined to de-escalate or retaliate, but not to escalate past whatever point will induce total nuclear war.6

Counter Argument/Conclusion

One might argue that in the event of a limited nuclear war, the U.S. should overtly advertise its limited ambitions. But leaders should also consider the effect that purposely advertising no intention to threaten a regime’s survival may have on the tactical objective itself. If the adversary becomes confident that the United States will not pursue regime change, it is possible that any effect of traditional nuclear deterrence will be lost.7 In other words, the adversary now believes that it can take any conventional measures necessary to win the fight and abandons any restraint it may have been exercising due to the fear of a potentially devastating U.S response. This is a significant challenge to the concept of pre-emptively guaranteeing the intentions of the United States, and it might be beneficial to not declare such intentions until circumstances have become dire enough that nuclear weapons are being considered.

Ultimately, the determining factor in “keeping the limited nuclear war limited” will be whether the authoritarian regime’s legitimacy is at stake. If the United States can effectively signal that any nuclear strikes are purely designed to win a tactical contest and not to effect regime change, it is rational to assume that a nuclear war would not go total. Perhaps this is wishful thinking, but even in the drastic case of nuclear use, it is crucial that rational actors actively take steps to prevent such a scenario. Given the number of potential hot spots that may necessitate a limited nuclear strike against autocratic regimes (South and East China Seas, the Baltics, and the Korean peninsula to name a few) policymakers and operators of our nuclear arsenal must take stock of the potential for escalation in the event these regimes feel threatened. Further research and consideration towards individual authoritarian regimes could help illuminate the specific actions that authoritarians may take when their legitimacy is threatened.

  1. This question most notably came up at the 2017 STRATCOM Deterrence Symposium ( Multiple presenters were extremely concerned about what the U.S. and NATO reaction would (or should) be to an overwhelming Chinese conventional war or a Russian “escalate to de-escalate” nuclear strike. Surprisingly, some more hawkish government-affiliated briefers wanted “first-use” policy explicitly defined in the new Nuclear Posture Review.
  2. This theory needs further study, but the behavior of modern authoritarians such as Putin suggests that they cannot justify their existence without some sort of existential threat. Peace and more freedoms would likely invite threats to their grip on power. Other regimes such as China and Saudi Arabia offer stability or massive social spending as a means of pacification, but they have not been immune to growing demands for freedom. The latter in particular has been able to keep its population relatively wealthy, but reforms are still being demanded.
  3. Robert Powell refers to these type of circumstances as “Nature”, i.e., things that happen in a game theory context that are not necessarily within the wishes of either rational party. See the following article for a brief explanation: Roeder, O. (2017, Sept 6). How to Win a Nuclear Standoff. Retrieved from fivethirtyeight:
  4. The “de-escalate, retaliate, escalate” options are adapted from Powell’s escalation ladder. This paper represents a new theory as to what variables affect those options, specifically with regard to authoritarian regimes. See: Powell, R. (1990). The Strategy of Limited Retaliation. In N. D. Theory, Nuclear Deterrence Theory (pp. 148-173). Cambridge: Cambridge University Press.
  5. “Escalation” may not necessarily include a nuclear strike, although there are few conventional options that could rival a nuclear attack. For example, however, if North Korea responded to a tactical nuclear strike on one of its isolated nuclear testing facilities with the conventional destruction of Seoul via artillery, this would irrationally invite regime change.
  6. In Powell’s model, each party always has the opportunity to choose total nuclear war and risk assured destruction. The more escalation that takes place, the less irrational such a decision becomes.
  7. In Thomas Schelling’s example, deterrence is like playing a game where two people are chained together at the edge of the cliff. The goal of each chainee is to scare the other into believing that one is willing to take the plunge, effectively deterring the other. However, if anti-regime change messages are too effective, it’s possible that you’ve just given your chained opponent permission to punch you repeatedly in the face, while assuring him you won’t take the leap.