What a Secret Cold War Game of Nuclear Hide-and-Seek Teaches Us About North Korean Verification

Some 50 years ago, military and government personnel spread out over an Army base in the American Southwest, looking for evidence of nuclear weapons: telltale cables, connectors, and other debris left over from a clandestine atomic test. The hunt was part of a little-known project designed to see if on-site verification of an arms control agreement was technically feasible. Code name: Cloud Gap.

This week, as U.S. President Donald Trump tries in a dramatic summit in Singapore to persuade North Korean leader Kim Jong Un to give up his country’s nuclear weapons, one of the central questions will be how to make sure the notoriously secretive regime won’t cheat. In the 1960s, the U.S. government faced a similar question as it weighed an arms control agreement with the Soviet Union.

The answer, some arms control officials hoped, could be found in Cloud Gap.

In its simplest conception, Cloud Gap was an attempt to see if inspectors could find evidence of an arms control violation in a given area. The project, started in 1963 and run out of a small office near the White House, included field exercises at U.S. military bases, where a red team playing the role of the Soviets would hide objects associated with a clandestine nuclear test, and a competing team would try to find them.

It was a kind of spy world treasure hunt.

The test, of course, presupposed an arms control agreement that allowed for reasonably open, on-site inspections, something the Soviet Union resisted during the Cold War and North Korea would likely resist today.

But the program, jointly run by the Defense Department and the Arms Control and Disarmament Agency, drew mostly ridicule among Washington’s nuclear experts. Robert Frosch, then a senior official in charge of nuclear monitoring research at the Pentagon’s Advanced Research Projects Agency, mocked the entire approach.

“I remember seeing documents and promptly dubbing it Loud Crap,” he said in an interview several years ago. “It was full of questions like, ‘You’ve got this city and there’s a surreptitious movement of tanks and artillery. … Will observers with binoculars know it?’ I’d been in lots of cities. If you were in the right part of the city, yes. If you were in the wrong part of the city, then no.”

For Frosch, the answer to the question of whether such on-site verification would work was obvious. “It all seemed a bit bizarre to me.”

Frosch’s concerns appear to have been at least partially borne out. A report written after one Cloud Gap exercise in 1967 reached the conclusion that inspectors would have difficulties distinguishing between the debris from real weapons and decoys.

“Inspectors’ abilities to discriminate between bona fide and fake nuclear weapons is poor,” the report said.

Stephen Lukasik, who also worked for the Advanced Research Projects Agency at the time, was equally skeptical. “What finished me was when one exercise failed to find a yellow school bus,” Lukasik recalled in an email. “It was fundamentally stupid: the area was too large and the likelihood of being given free rein in the USSR was vanishingly small.”

Satellites and other technological methods of monitoring would be far more reliable for detecting nuclear work, Lukasik decided.

If technical concerns didn’t finish off the program, then bad luck did.

In April 1967, in the midst of a Cloud Gap exercise, a helicopter carrying out a mock inspection crashed in Arizona’s Santa Rita Mountains, killing several crewmembers. Shortly after the crash, the Johnson administration decided to end the program, leaving unanswered the question of on-site inspections’ effectiveness.

While the focus has now shifted from the Soviet Union to North Korea, the question remains relevant: Can on-site inspections ensure another country has given up its nuclear weapons?

“It is an interesting episode, worth revisiting,” Steven Aftergood of the Federation of American Scientists wrote in an email when asked about Cloud Gap’s relevance to North Korea. “A successor program today would have more sophisticated tools available both for detecting nuclear materials and for protecting sensitive design information.”

While Cloud Gap’s existence was known, its findings were kept secret for decades. Many of the records were released just a few years ago in response to a Freedom of Information Act request filed by Aftergood.

Former Defense Secretary Harold Brown, who served in a senior position at the Pentagon during the time Cloud Gap was conducted, believes that the central difficulty of verification remains unchanged, even as the methods for on-site inspections have improved.

“The question was always when (and especially how and whether) access would be allowed,” he wrote in an email. “That is still the question.”

Humza Jilani contributed research to this article.