By the mid-1960s, both the United States and the Soviet Union had built nuclear arsenals large enough to physically destroy each other twice over within thirty minutes of a launch order. The strategic doctrine that emerged from this fact — Mutually Assured Destruction, with the unimprovable acronym MAD — held that nuclear war between the superpowers was unwinnable, and that the only stable equilibrium was for both sides to know with certainty that any first strike would result in their own annihilation. The phrase was popularized by analysts at the RAND Corporation and the Pentagon under Robert McNamara, who turned the grim arithmetic of second-strike capability into formal policy.
The doctrine has the structure of a paradox: peace is preserved by credibly threatening apocalypse. Stability requires that retaliatory forces be invulnerable to a first strike — hence the nuclear triad: submarine-launched missiles hidden in the oceans, road- or silo-based land missiles, and bombers once kept on permanent airborne alert. It also requires that defense itself be renounced, the logic behind the 1972 ABM Treaty: a shield that worked would let one side strike first without fear of reprisal, breaking the symmetry that kept the peace. And it requires that the other side believe you would actually use them, which means cultivating a kind of strategic insanity at the highest level — Nixon even courted it deliberately, his 'madman theory' — and automating retaliation enough that no decapitating strike could silence it, as the Soviets did with their semi-automatic Perimeter system. The Cold War was, against all expectation, the longest stretch without war between great powers in modern European history. Whether this was because of MAD or despite MAD is among the most important unresolved arguments in political science. The optimistic reading is that nuclear weapons made great-power war irrational and obsolete. The pessimistic reading — pressed by close calls like 1962, the 1983 Able Archer scare that Soviet leaders genuinely mistook for war preparation, and false-alarm incidents like Stanislav Petrov's 1983 refusal to report a faulty launch warning — is that we were lucky for forty-five years, and luck is not a strategy.
Nine states now have nuclear weapons; perhaps a dozen more could build them within a year. Russia rattling its arsenal over Ukraine, North Korea under no constraint, an Iran that may yet decide, an India and Pakistan that already treat them as coercive tools — the assumptions of MAD do not generalize cleanly to a multipolar nuclear world, where deterrence must hold across many asymmetric, mutually distrustful pairs at once. This may be the most consequential under-priced risk on the planet.