Inspecting nuclear warheads like they were passwords

Nuclear weapon inspectors have a weighty but tricky job. An inspecting state relies on them to verify if a weapon is a nuclear warhead, but the state whose weapons are being inspected doesn’t want to divulge too much information about the weapon’s design or performance. As David Cliff, a researcher at the Verification Research, Training and Information Center, London writes,

In warhead dismantlement, the objective needs to be to gain as much confidence through agreed verification measures as possible, thereby minimizing the extent to which trust [between the two states] will need to become a factor… In fact, as a means of building trust and confidence between states, dismantlement is of limited value unless it occurs in a transparent and verifiable manner.

(Emphasis mine)

So, on the one hand, transparency is needed to ensure the number of warheads have been reduced. On the other, secrecy is necessary to keep the warheads from reaching the hands of potential adversaries, not to mention to ensure deterrence. Methods to measure sensitive information often include safeguards to protect it, adding another layer of liability.

To simplify this process, researchers from the USA and UK have developed a new technique to verify warheads without needing any sensitive information about them – thus eliminating the need for them to be made available in the first place. They propose to bombard a supposed warhead with neutrons, then using a detector to check the properties of the particles that have passed through. Next, an actual known warhead is subjected to the same profiling.

Then, an inspector randomly chooses to use each detector on other warheads that need to be inspected. Over multiple tests, the detector will be able to check with increasing likelihood if a warhead is genuine or not by comparing it to previous tests. Crucially, the inspector will not have access to any parameters of the comparison but only if a ‘Yes’ or a ‘No’ has been signalled.

A paper describing this ‘zero-knowledge protocol’ appeared in Nature on June 26, in which the researchers argue,

This technique will reveal no information about the composition or design of nuclear weapons when only true warheads are submitted for authentication, and so does not require an engineered information barrier.

To assist in their analysis, the team used the unclassified British Test Object (BTO), which “does not contain special or other nuclear materials, but is used to develop and calibrate imaging systems for diagnostic analysis of nuclear weapons”. It consists of concentric rings of polystyrene, tungsten, aluminum, graphite and steel. Over multiple tests (i.e. simulations) on the BTO, the team then estimates the number of tests needed to reliably detect increasingly serious defects, finding 5,000 and 32,000 to be sufficient to detect the most serious ones.