Thursday, November 1, 2007

The Search for Scorpion

On May 27, 1968 USS Scorpion was reported missing with ninety-nine men on board. Nobody had any idea where Scorpion was or what had happened to her. All they knew was that the 3,500 ton nuclear attack submarine was due back in Norfolk, VA and had failed to arrive.

Dr. John Craven, the Chief Scientist of the U.S. Navy's Special Projects Division had been instrumental in locating a missing Hydrogen Bomb off the coast of Spain on January 17, 1966 and he was to play a key role in locating Scorpion.

Craven and his team of mathematical consultants launched a search that would take so many twists and leave him so at odds with the rest of the navy that he himself would begin to wonder whether he had indeed gone mad.

Craven and his team together with Wilton Hardy of the Naval Research laboratory and his team of acoustic experts examined the acoustic data. Hardy found it first. Right on Scorpion's track was an explosion strong enough to tear through a steel hull and 91 seconds later there was a series of much louder blasts as the submarine began imploding as it sank to the bottom.

The site of the first explosion – codenamed Point Oscar marked where the search would begin. The water at Point Oscar was 2 miles deep. The Scorpion would have stopped imploding about 7,000 feet before she hit bottom, cutting of the acoustic trail. Depending on how fast she had been traveling, and in what direction, and depending on the force of the implosion and the position of her stern planes as she fell, she could have been thrown miles further. All that meant that the submarine could be anywhere within a 20-mile-wide circle, leaving a vast unknown universe to search.

Craven began digging for more evidence. He set about trying to map each implosion in the hope that he could figure out how far scorpion had traveled before the final sounds of her loss subsided. Craven's map showed that Scorpion had not been traveling west toward Norfolk during her final moments. Instead Craven's calculation surprisingly showed that the submarine had been moving east back toward the Mediterranean. Craven asked several captains and admirals "what would make a submarine go in the wrong direction". Each time he got the same answer. A submarine turns around 180 degrees when a torpedo activates when it is still on board. The boat turns because that triggers fail-safe devices on the torpedo shutting it down.

The only problem was that nobody of any rank from the chief of naval operations on down thought Craven could be right. Craven persisted, he began to mathematically construct a map of the ocean bottom, using Bayes' Theorem of subjective probability to draw on the knowledge that even experts are not consciously aware they have. After Craven explained to the Navy brass that he was going to use a system of Las Vegas-style bets to factor the value of hunch into his data, some of the operational commanders were convinced that he had gone completely over the edge.

Undeterred Craven asked a group of submarine and salvage experts to bet on the probability of each of the different scenarios being considered to explain scorpions loss. Once the bets were completed Craven sat down to draw a probability map. The calculations concluded that the scorpion was east of Point Oscar, 400 miles from the Azores on the edge of the Sargasso Sea.

Years later mathematicians would write a book based on their work with Craven entitled Theory of Optimal Search, the U.S. Coast Guard would adopt the method for search and rescue and the navy would use Craven's interpretation of Bayes' Theorem to locate sunken ordnance in the Suez Canal.

Yet in the Scorpion search naval officers just shook their heads and continued to look to the west of Point Oscar. Months passed by with no trace of Scorpion. By October with the weather worsening Craven finally persuaded the navy to look east of Point Oscar. On October 29, five months after she had been reported missing Scorpion was found within 220 yds of where Craven, his mathematicians and a group of experts betting for a bottle of Scotch had said she would be.

The above is an abridged version of a chapter in the book Blind Mans's Bluff: The Untold Story of American Submarine Espionage. The book is a great read but one might wonder how it relates to our business.

There are many good lessons to be learned from the book but what is particularly interesting is the technique Craven used to draw on the knowledge that even experts are not consciously aware they have. Not only that but it is impressive the quality of the estimates. We deal with uncertainty in our line of business. Most notably in risk management.

When we implemented our agile risk management technique we used a modified Delphi technique to increase our confidence in a series of subjective measures. However, the word subjective always bothered me. At the time intuitively I knew that behind the scenes of every expert judgment was a series of decisions based on real experiences through the years and just because those data points were not discussed or written down did not mean they did not exist. They perhaps just flashed through someone's mind quicker than a value remains in a register on a chip but they were there. I really did not have any science to back up this theory but then I read Blind Mans Bluff and there it was, a real-life example that worked and was backed up by science. I think if we had known about Craven's work and Bayes' Theorem at the time we might have been able to make a much stronger case for the validity of our approach. An approach that at the time yielded some alarmingly high numbers yet reflecting on them now, our team of experts were also spot-on.