USS Scorpion

USS Scorpion was a U.S. Navy submarine lost in May 1968 on its return to Newport News from a tour of duty in the North Atlantic. Its disappearance became a striking test of whether scattered judgments could locate what no individual expert could: John Craven assembled a group to estimate the submarine’s fate, and the resulting collective estimate came within 220 yards of its final location.

Disappearance and Search

The Navy knew Scorpion’s last reported position but did not know what had happened after its last radio contact or how far it might have traveled. The initial search area was a circle twenty miles wide and thousands of feet deep, a vast volume in which to look for a submarine whose course and fate were uncertain. The apparent alternative was to ask a handful of leading submarine and ocean-current experts where they thought it lay. Craven instead sought a way to combine many separate judgments.

Building a Collective Estimate

Craven devised several possible explanations for what had happened to Scorpion and assembled men with varied expertise, including mathematicians, submarine specialists, and salvage workers. Rather than have them confer and settle on a shared answer, he asked each man to estimate the likelihood of each scenario. Their guesses took the form of wagers, with bottles of Chivas Regal as prizes. The bets addressed distinct questions: why the submarine had run into trouble, how fast it traveled toward the ocean floor, and how steeply it descended.

From Wagers to Location

No single guess could locate the submarine. Craven’s method depended on treating the wagers as pieces of a composite account of Scorpion’s final moments and using Bayes’s theorem to estimate its last location. The theorem updates prior expectations in light of new information; here, it supplied a way to combine estimates about separate uncertainties into a collective judgment.

The result did not match any individual member’s chosen location. No one person had held the complete picture that Craven constructed from the group’s answers. The estimate was therefore collective in a substantive sense, not simply the judgment of the group’s smartest member. Five months after the disappearance, a Navy ship found Scorpion just 220 yards from the predicted location.

Significance of the Loss

The search’s achievement was especially striking because the evidence available to the group consisted of tiny scraps. The men did not know why Scorpion sank, how fast it moved, or how steeply it fell. Yet by keeping their judgments separate and combining them, the group produced a location that none had independently selected. The submarine’s loss thus serves as the introduction’s complex counterpart to a simple group estimation problem: very different in scale, but illustrating the same principle of collective judgment.