COME BACK, ADMIRAL POINDEXTER, ALL IS FORGIVEN:

In 2011, a former Pentagon strategist named Phillip Karber who was teaching at Georgetown University asked his students to study the Chinese tunnel system known as the "underground great wall." The tunnel's existence was well-known, but its purpose was not. Karber's students turned to commercial imagery, blogs, military journals, even a fictional Chinese television drama to get answers. They concluded the tunnels were probably being used to hide 3,000 nuclear weapons. This was an astronomical number, about 10 times higher than declassified intelligence estimates and other forecasts of China's nuclear arsenal.

The shocking findings were featured in the Washington Post, circulated among top officials in the Pentagon, and led to a congressional hearing. They were also incorrect.

Experts quickly found egregious errors in the study. A Harvard researcher found that Karber's students based the 3,000 weapon number on an American intelligence projection from the 1960s, assumed it was accurate, and then just kept adding weapons at a constant rate of growth. Apparently, they did not take seriously more recent declassified intelligence estimates that China probably had no more than 200 to 300 warheads. And Karber's estimates for the amount of plutonium needed for the weapons were based on sketchy sources using even sketchier ones: The study cited Chinese blog posts that were based on a plagiarized grad school essay from 1996, which in turn relied on a single anonymous post on the site Usenet. The plutonium sourcing was "so wildly incompetent as to invite laughter," wrote nonproliferation expert Jeffrey Lewis.

Phillip Karber speaking on a screen at the U.N. Security Council Arria Meeting on Crimea.

Former Pentagon strategist Phillip Karber, whose Georgetown students shockingly -- and wrongly -- concluded that China was hiding thousands of nuclear warheads. | Estonian Foreign Ministry/Flickr

This is the radical new world of open-source intelligence -- where crises move faster, information is everywhere and anyone can play. Intelligence isn't just for governments anymore, thanks to three major trends over the past several years: the proliferation of commercial satellites, the explosion of Internet connectivity and open-source information available online, and advances in automated analytics like machine learning. These changes have touched every part of the intelligence landscape. In particular, they've given rise to a host of non-governmental detectives who track some of the most serious and secret dangers of all: nuclear weapons.

The world of open-source nuclear sleuthing is wide open to anyone with an Internet connection. It draws people with a grab bag of backgrounds, capabilities, motives and incentives -- from hobbyists to physicists, truth seekers to conspiracy peddlers, profiteers, volunteers and everyone in between. Many are former government officials with years in the field, but others are amateurs with little or no experience. There are no formal training programs, ethical guidelines or quality control processes. And errors can go viral; nobody loses their job for making a mistake.

The open-source revolution has been lauded for disrupting and democratizing the secretive world of intelligence. There is no doubt that open-source intelligence is invaluable and that spy agencies must find new ways of harnessing its insights. But the news is not all good. Citizen-detectives also generate risks. From the most obvious risk of getting it wrong, to harder-to-see downsides like derailing diplomatic negotiations by publicly revealing sensitive findings, the U.S. intelligence community needs to pay attention to the potential dangers of open-source intelligence as it adapts its spycraft to the digital age.

Tweet   @brothersjudd