Two suspects are interrogated separately, with no possibility of speaking to one another. In this setup, several outcomes can unfold: if one of them confesses and the other does not, the first goes free while the partner receives the maximum sentence; if both confess, both are condemned to an intermediate penalty. And if neither confesses, both will be sentenced to the minimum penalty. The logic suggests that, in the absence of mutual trust and even though cooperation is the most suitable option, the rational move is to betray before being betrayed. The prisoner’s dilemma illustrates the logic that has sustained the nuclear arms race for eighty years.
“The prisoner’s dilemma illustrates the logic that has sustained the nuclear arms race for eighty years”
After World War II and, more specifically, after the Cold War, there were numerous attempts to curb the nuclear arms race in the face of the exponential risk arising from technological development. During this period, Daniel Ellsberg, an American defense analyst and disarmament activist, exposed in his work The Doomsday Machine: Confessions of a Nuclear War Planner the doctrinal distinction between First Strike and First Use.
The First Strike refers to a large-scale attempt to disarm the opponent in advance of a nuclear exchange. The main objective is to destroy air bases, missile silos, submarine bases, and command centers to nullify or degrade their ability to respond. The First Use, on the other hand, refers to the decision to introduce nuclear weapons into a conflict for the first time, either to complement a non-nuclear defense or to reverse a failed military operation.
In Ellsberg’s reading, this distinction reflects a contradiction within the doctrine of deterrence based on retaliation, since First Strike is treated as a real option; that is, betrayal is prioritized over cooperation. Not out of malice, but as a matter of survival logic.
Translated to the international arena, the prisoner’s dilemma takes on colossal and truly complex dimensions. Jonathan Cave, in Introduction to Game Theory (1987), presents the dilemma through two strategies: cooperation (Helpful, H) or betrayal (Greedy, G).
If both States decide to disarm (mutual cooperation), the outcome is as good as it gets: global risk diminishes and military spending falls. In a second scenario, if one State chooses to modernize its nuclear arsenal and preserve the option for a First Strike (does not cooperate), while the other disarms unilaterally (cooperates), the second finds itself in a critically vulnerable position, exposed to a decisive attack. In the third scenario, if both decide not to disarm (do not cooperate), an equilibrium of subsistence is reached that is worse than shared cooperation, but better than unilateral defenselessness.
Cave’s analysis shows that, in a single interaction, the dominant strategy for both actors is betrayal. Yet, in a framework of indefinitely repeated play, cooperation becomes more plausible. Not out of generosity, but due to the deterrent incentive of future retaliation. Paradoxically, this reveals a central risk in the architecture of nuclear deterrence: the greater the conviction of an escalation toward the end of the world, the lower the incentive to cooperate. Cooperation during the Cold War, through arms reduction agreements like START I, did not crystallize in moments of panic, such as the Cuban Missile Crisis, but during periods of détente when both sides contemplated a shared horizon.
“There is a central risk in the architecture of nuclear deterrence: the greater the conviction of an escalation toward the end of the world, the lower the incentive to cooperate”
To make an international agreement binding and effective, good will alone is not enough; without real incentives, it becomes far more convenient to break what was agreed. Cave proposes altering the very payoff matrix by introducing an extra cost in case of betrayal, which he calls “hostage-taking institutions.” In the nuclear context, the ultimate expression of this guarantee is Mutually Assured Destruction (MAD), which Cave himself cites as an example. Nevertheless, the author warns of a substantial risk that today can be read as one of the keys to the weakening of the non-proliferation and disarmament regime: a hostage system works only if these hostages can be verified as real. Without a mechanism able to confirm, independently and reliably, that disarmament and non-proliferation commitments are truly met, any normative advance runs the risk of standing on promises that neither party can verify nor, therefore, trust.
This is the case of the Treaty on the Non-Proliferation of Nuclear Weapons, opened for signature in 1968 and conceived as an agreement aimed at overcoming this paradigm. On the one hand, non-nuclear-armed states committed not to manufacture or acquire them, while the Treaty acknowledges everyone’s right to the peaceful use of nuclear energy; on the other, the parties undertook a legal obligation to negotiate in good faith effective measures related to nuclear disarmament. Today, the obligation to disarm lacks real enforceable efficacy, while the architecture of the international non-proliferation regime faces unprecedented erosion.
“At present, the obligation to disarm lacks real enforceable effectiveness, while the architecture of the international non-proliferation regime faces unprecedented erosion”
The deterioration of the regulatory framework is evidenced by Russia’s suspension of the New START treaty in 2023 and its expiration on February 5, 2026, as well as by the emergence of cutting-edge technologies such as artificial intelligence in systems related to nuclear command and the spread of hypersonic missiles, elements that pose challenges not fully covered by existing control frameworks and strain the guarantees of cooperation.
It is a paradox. The more imminent the threat to national security is perceived, the weaker the incentive to disarm through a treaty. Game theory applied to the analysis of international law teaches that peace is not sustained by good will nor by fear, but by the guaranteed certainty that the other side also complies with what has been agreed. Without independent verification mechanisms with evidentiary value, any normative advance in disarmament becomes mere parchment.
