Chuck Bednar for redOrbit.com – Your Universe Online
Boredom can be a dangerous thing in the hands of a researcher – just ask Randy Olson, a grad student at the Michigan State University High-Performance Computing Center who decided to develop an algorithm designed to crack the popular Where’s Waldo? series of books.
For those who aren’t familiar with ‘Where’s Waldo?’ the basic concept is simple: it’s basically a hide-and-seek game where readers have to find the titular character, who is always dressed in red and white striped clothes and resembles a cross between a hipster and a candy cane, as quickly as possible. The books are illustrated by Martin Handford, and the first was released in 1986.
Enter Olson, who according to a recent blog entry found himself unexpectedly snowed in one weekend. While searching for something to occupy his time, he came across a 2013 Slate article claiming to have developed a foolproof strategy for finding Waldo. After reading the story, and believing that he could improve upon their method, Olson set to work on a new algorithm.
Using a chart provided by the original article, he was able to locate all 68 of Waldo’s coordinates in the first seven primary editions of the books. He then performed a kernel density estimation of those points to detect several trends: notably, that Waldo is almost never in the top left corner, is rarely located on the edges, and is never located on the bottom of the right page.
With that information, he set out as a way to check every possible location as quickly as possible and without backtracking. So he took at list of the 68 points where Waldo could be, and arranged them in the order he intended to visit them. This meant trying every possible arrangements of the points to find the one with the least distance traveled. Doing so was no easy task.
“Those 68 points can be arranged in ~2.48 x 1096 possible ways,” Olson said. “To provide some context, that’s more possible arrangements than the number of atoms in the universe. That’s so many possible arrangements that even if finding Waldo became an international priority and the world banded together to dedicate the 8.25 million computing cores from the world’s 10 largest supercomputers to the job, it would still take ~9.53 x 1077 years – about 6.35 x 1067x longer than the universe has existed – to exhaustively evaluate all possible combinations.”
“Thankfully, there are plenty of smarter methods for approximating the optimal search path for finding Waldo,” he added. What Olson did was use a genetic algorithm to visualize the optimum solution over time. Genetic algorithms constantly make slight changes to a good solution in order to try and find a better one until there are simply no more improvements to make.
After running the algorithm for five minutes, Olson said found the best possible solution. Based on his algorithm’s findings, the best place to start is the bottom of the left page. If he’s not there, the reader should then check the upper quarter of the right page, followed by the bottom right half of the right page, and finally the bottom left half of the right page as a last resort.
Of course, as Mashable points out, using science to beat a kids game kind of takes all the fun out of it. “If you want to keep you childhood nostalgia intact, we recommend you abstain from using science in your search,” the website said. “However, if you get stuck to the point where you want to pull your hair out, Olson’s method will definitely save you a lot of stress.”
—–
Follow redOrbit on Twitter, Facebook, Google+, Instagram and Pinterest.
Comments