Proving the Negative: Three Warnings from Arms Control for the Age of AI

At the center of the 1963 negotiations was not whether verification mattered, but how much on-site access each country should receive. The United States wanted seven inspections each year. The Soviet Union, led by Nikita Khrushchev, offered three. The disagreement reflected a deeper struggle between confidence and sovereignty. Each side wanted evidence that the other was obeying a comprehensive test ban, but neither wanted inspections to become a tool for interference or espionage.

The gap mattered because seismic evidence was imperfect. An underground nuclear explosion could resemble an earthquake in seismic records. Inspectors could visit suspicious sites, collect information, and increase confidence about what had happened. Yet the two governments also disagreed about inspection procedures and treaty language, so the number was part of a larger verification dispute.

The negotiations show why arms-control treaties need both clear rules and trusted monitoring. A compromise can be blocked when verification seems too weak to reassure one side or too intrusive to satisfy the other. Later agreements continued this balance, but comprehensive verification remains politically difficult.

Nikita Khrushchev proposed three on-site inspections each year, while Washington asked for seven. The numerical gap was four inspections annually. That may sound small, but it represented a major disagreement over how much direct access was needed to verify a comprehensive nuclear test ban. More inspections could make suspicious seismic events easier to investigate. Fewer inspections could limit foreign access and reduce fears of abuse.

The key mechanism was uncertainty. Seismology could detect underground disturbances, but it could not always identify their cause. An earthquake and an underground nuclear explosion might produce similar signals. An on-site inspection could provide additional evidence, helping inspectors judge whether a treaty violation had occurred. Thus, the number of inspections affected the credibility of the entire ban, not merely administrative planning.

The 1963 dispute illustrates a continuing arms-control problem: verification must be strong enough to reassure, yet acceptable enough to win consent. Modern monitoring uses several technologies, but disagreements over access and trust still shape treaty politics.

A comprehensive nuclear test ban is an agreement forbidding nuclear weapon test explosions in all environments. Its purpose is to stop countries from using explosions to develop, improve, or demonstrate nuclear weapons. Unlike a limited ban, it would cover tests underground as well as in the atmosphere, underwater, and outer space. The central idea is completeness: no testing route should remain available.

This scope matters because underground tests can be hidden more easily than atmospheric explosions. They still produce seismic signals, but those signals can sometimes resemble earthquakes. A comprehensive ban therefore depends on monitoring and, when necessary, inspections. Verification does not create the prohibition; it helps countries determine whether the prohibition is being respected.

In current arms-control practice, the Comprehensive Nuclear-Test-Ban Treaty seeks this goal but has not entered into force. Its monitoring system operates internationally, while the treaty’s legal status remains incomplete. The continuing challenge is combining a universal rule with confidence that violations would be detected.

Seismology measures waves traveling through Earth after a sudden release of energy. Both earthquakes and underground nuclear explosions create such waves. If their size, depth, geology, or location produce similar signals, the first seismic record may not clearly reveal which event occurred. This is why the article says seismology could not always distinguish an underground test from an earthquake.

The key mechanism is signal overlap. Scientists compare features such as wave types, strength, depth, and timing, but local rock conditions can alter the signals. A small explosion may look ambiguous, and an earthquake can occur near a suspected testing area. An on-site inspection can add physical and environmental evidence that seismic measurements alone cannot provide.

Today, seismic networks are much more extensive and analytical methods are more advanced, but ambiguity has not disappeared. Effective verification therefore combines seismic data with other sensors and, where treaties permit, inspections. The broader lesson is that detecting an event is easier than confidently identifying it.

When countries cannot reliably verify compliance, they cannot easily tell restraint from concealment. That uncertainty can weaken confidence in a nuclear test ban. Governments may suspect that another country is testing secretly, even without decisive proof. They may then resist signing a treaty, reject compromises, or seek stronger military and intelligence safeguards.

The 1963 negotiations show the mechanism clearly. Seismology could not always distinguish an underground explosion from an earthquake. Inspections offered a way to investigate ambiguous events, but the United States and Soviet Union disagreed over how many were acceptable. Too few could leave violations hidden; too many could seem intrusive or exploitable. Verification therefore became part of the political dispute itself.

The consequence today is a continuing demand for layered monitoring, transparent procedures, and reliable access. No system removes every uncertainty. Strong systems make violations more detectable and deception riskier, helping treaties survive political pressure. Weak systems can turn even lawful activity into a source of suspicion.

Besides seismic data, countries can monitor radioactive materials, atmospheric pressure waves, ocean sounds, and visible changes on the ground. Radionuclide stations may detect radioactive particles or noble gases released by a test. Infrasound sensors listen for low-frequency pressure waves from atmospheric explosions. Hydroacoustic sensors monitor sound traveling through oceans, while satellites can observe flashes, ground disturbance, heat, or construction activity.

Each method works through a different mechanism. Seismic sensors detect energy moving through Earth. Radionuclide systems detect radioactive traces. Infrasound and hydroacoustics detect pressure or sound waves in air and water. Satellites provide images and other remote observations. Because each method has limits, matching evidence across systems can help distinguish a nuclear test from a natural event or ordinary activity.

Modern verification relies on this layered approach, especially through the International Monitoring System associated with the Comprehensive Nuclear-Test-Ban Treaty. Technology can improve detection, but access, data sharing, and political cooperation remain essential. Multiple methods reduce uncertainty without making any single sensor decisive.

Proving that something happened usually requires one convincing piece of evidence: a signal, sample, image, or witness. Proving that it did not happen is harder because the event may have been concealed, missed, or too small to detect. Inspectors must show that monitoring covered the relevant places and times, that sensors worked, and that no plausible evidence escaped observation.

For a secret nuclear test, the mechanism is a coverage problem. A country could conduct an activity in a remote area, disguise its signals, or exploit limits in detection systems. Even if monitoring finds nothing, that result may mean no test occurred—or simply that the system lacked sensitivity or access. This is why the 1963 debate over inspection numbers mattered: inspections could reduce uncertainty, but they could not guarantee perfect knowledge.

Modern verification seeks high confidence, not magical certainty. It combines sensors, inspections, data exchanges, and repeated observation. The forward implication is important: treaties are strongest when violations are likely to be detected, even though proving complete absence remains impossible in practice.