I had the privilege to be a judge for Business Insurance's Risk Manager of the Year awards, presented at the Risk & Insurance Management Society Inc.'s annual conference. As I weighed the submissions, I realized serious risk managers were careful to avoid risk management processes that depend upon predicting future events. That encouraged me to try an experiment.

The experiment

The exercise involved 22 MBA candidates at St. Peter's College who were taking a course in enterprise risk management. They were divided into four teams and shown a slide with the following words: “Two Eskimos sitting in a kayak were chilly, so they lit a fire in the craft. Unsurprisingly, it sank. What does this prove?” After two minutes, the teams gave answers. None matched the pun, which was: “You can't have your kayak and heat it too.” Recognizing the play on the pun “Have your cake and eat it too,” the group prepared for questions with a play on words.

The instructor advised the students that they would be given five similar phrases. Each individual wrote down a forecast of the likelihood that his or her group would match an answer taken from an Internet posting. For each question, the teams had two minutes to develop the answer. Then, they received immediate feedback from the instructor.

c Question No. 1: “A Buddhist went to the dentist and learned that he needed a root canal. He refused Novocain. What was he trying to do?” Two answers made some sense: “No pain, no gain” and “Getting to the root of the problem.” The Internet answer: “The Buddhist wanted to transcend dental medication.”

c Question No. 2: “A hotel hosted a convention of chess masters. A group stood in the lobby bragging about their winning games. The hotel manager asked them to disperse. Why did he do that?” No answer was close to: “The hotel manager could not stand chess nuts boasting in an open foyer.”

c Question No. 3: “A woman had identical twins and gave them up for adoption. One went to Egypt and was named Ahmal. The other went to Spain and was named Juan. Years later, Juan sent a picture of himself to his birth mother, who told her friend that she wished she also had a picture of Ahmal. What did the friend say?” The answer was: “They're twins! Once you've seen Juan, you've seen Ahmal.” One team got it.

c Question No. 4: “Two friars opened up a florist shop. A rival florist hired a vicious thug, Hugh McDonald, to "persuade' them to close. Hugh beat up the friars and they closed the shop. What does that prove?” Although all the teams saw that “Hugh” needed to become “you,” no group came close to “Only Hugh can prevent florist friars.”

c Question No. 5: This question was placed in the last position because it was the most difficult. “Mahatma Gandhi walked barefoot. He had heavy calluses on his feet. He ate very little so he was frail. From a poor diet, he had bad breath. What did these things make him?” After two minutes, all the groups were completely stymied. Then, the class was shown the same slide with two added words and five words displayed in bold print: “Mahatma Gandhi (Indian mystic) walked barefoot. He had heavy calluses on his feet. He ate very little so he was frail. From a poor diet, he had bad breath. What did these things make him?”

Within seconds, one student blurted out, “supercalluses,” followed by another who said, “frail mystic,” and another said “halitosis.” In a classwide collaboration, everyone recognized “Supercalifragilisticexpialidocious,” a song title from the musical “Mary Poppins.” The class quickly got most of “Mahatma Gandhi was widely known as a supercallused, fragile mystic hexed by halitosis.”

Lessons learned

The students displayed overconfidence in their ability to match the Internet answers. Overall, they estimated a 51% chance of matching all five answers. This overconfidence was expressed despite the obvious fact that no group came close to heating the kayak.

As they created answers and received feedback, responses improved significantly. Although only one answer matched exactly, several were quite on target: “Sometimes you can't twin Ahmal,” and, “You can't have Juan without a mother.”

Collaboration improved with each question. At first, the group members engaged in little discussion. By question No. 5, everyone was throwing out ideas.

Focusing groups on key points improved problem solving. When “mystic” was added and “calluses,” “frail” and “bad breath” were highlighted in bold on the final question, rapid-fire responses quickly uncovered the saying.

This experiment encourages us to beware of excessive confidence in our predictions, allow individuals to gain experience and collaborate as they deal with uncertainty, and seek clues that focus us on correct risk decisions. The results support the creation of a central risk function that searches out danger and opportunity, shares its findings so executives can collaborate and alerts people about the danger of overconfidence. These are some things to think about as we honor risk managers who already know these lessons.