Fred Hazelton wants you to defeat Evil Emperor Zurg, arch-nemesis of Buzz Lightyear.
But he doesn't want you to stand in line for hours, roasting in the Florida sun, waiting for the battle to begin. He wants you to ride Cinderella's Golden Carousel, too. But he doesn't want you to suffer a two-hour wait for a two-minute spin.
Hazelton, an Ottawa resident with a math degree from the University of Waterloo, spends much of his time crunching numbers for Statistics Canada. But for 15 hours each week, the 35-year-old statistician and father of two turns his attention southward, to Orlando, Fla.
Hazelton is the official statistician for the Unofficial Guide to Walt Disney World, a perennial bestseller. His expertise: predicting crowd levels. His mission: help visitors to the world's busiest recreational resort spend more time on their bums, less time on their feet.
"In the case of Disney World," he says, "the more you see and the less you wait, usually the happier you're going to be."
At age four, Hazelton visited Disney World for the first time. In 1998, while planning his first visit as an adult, he read an article about Unofficial Guide author Bob Sehlinger, who was looking for a Disney-loving statistician.
"Well," Hazelton thought. "That's me."
He contacted Sehlinger and was put in touch with Len Testa, an Unofficial Guide contributor. Testa, who has a Mickey Mouse tattoo on his right ankle and co-hosts a thrice-weekly Disney podcast, is the brains behind a computer algorithm that helps people get around Disney World with minimal delays. On a busy day, a Testa touring plan can save you more than four hours of standing in line.
It was a two-hour wait for The Great Movie Ride, in 1994, that motivated Testa to begin work on his touring-plan software.
"I thought that there had to be a better way of doing this," says Testa, a 40-year-old software developer for American Express in Greensboro, N.C.
The project became his masters thesis at North Carolina Agricultural and Technical University. Of course, determining the best route between multiple destinations isn't a novel pursuit. It's a well known challenge among mathematicians, who call it the Time Dependent Travelling Salesman Problem. For some organizations, such as the package delivery company FedEx, efficient travel is a top priority. Testa's primary difficulty was incorporating unique theme park constraints like hours of operation and attraction wait times, among other things: "When you plan a family's day in Disney World, you have to consider lunch."
Testa completed his masters in computer science in 2000, and within three years his touring plans were part of the Unofficial Guide. But the plans aren't perfect. A person wanting to complete 21 rides could do it many ways -- 51,090,942,171,709,440,000 ways, to be exact. The Magic Kingdom, Disney's most popular theme park, has nearly 50 attractions. Evaluating all possible paths to determine the best wouldn't be practical, no matter how powerful the computer.
Testa's tours -- which come in many versions, from a plan for preteen girls who don't like roller coasters to one for seniors who want to minimize walking -- are optimal to within two per cent.
That was good enough for Ed Waller, a Texas resident who in 2003 used a touring plan to do what many thought couldn't be done: visit every Magic Kingdom attraction in one day.
"I remember that day like I remember the birth of my daughter," says Testa. "It was validation."
Visitors to the Unofficial Guide's website (touringplans.com) can generate custom touring plans for any Disney park. Some 20,000 users have created plans, most of which are free on the website. But the software only works if Testa knows the approximate wait time at every attraction at any time on any day. Getting that data requires Hazelton's statistical know-how. And lots of legwork.
Testa, Hazelton and a team of volunteers spend several weeks a year collecting data at Disney World. When in Orlando, everyone repeats an assigned half-hour loop in one theme park until it closes, sometimes walking nearly 30 kilometres. Data is recorded in custom notebooks that won't smudge if it rains. (Testa tested them by standing under his sprinkler.)
Once the data is collected, Hazelton uses it to predict wait times for future dates, though he must also consider many other factors that affect crowd sizes, such as the season, special events, or extended hours. About 80 to 90 per cent of the time, his predictions are accurate to within eight minutes. Errors usually stem from unpredictable events like a ride breaking down.
"When we're wrong, it's because of something like the weather, which Fred can't yet control," says Testa. "But we're getting him to work on that."
-- Canwest News Service