When Jeopardy master Ken Jennings recounted his loss to computer Watson early this year, he referred to the game show’s questions as an “open domain of human knowledge — every subject imaginable.” But according to computer scientist Roger Craig, that conception is flawed. Using a self-built pattern recognition program to focus his studies on the most common topics, Craig has won six games, plus a Tournament of Champions, and taken the all-time record for money won in a single game.
Jeopardy contestant Roger Craig stomps competition with data mining
Roger Craig’s data mining study method has netted him a successful five-time run on Jeopardy, as well as a record for the highest-ever winnings from a single game.
Roger Craig’s data mining study method has netted him a successful five-time run on Jeopardy, as well as a record for the highest-ever winnings from a single game.
is a senior tech and policy editor focused on online platforms and free expression. Adi has covered virtual and augmented reality, the history of computing, and more for The Verge since 2011.
Unlike Watson, which relies on stored terabytes of general information, Craig analyzes a database of past Jeopardy questions, arranging them into groups and then looking for patterns and recurring questions. Once he knows which topics have a good chance of coming up, he allocates his study time accordingly — instead of trying to learn every capital in the world, Craig focuses on the most statistically popular. For him, it’s not a question of knowing everything, but of knowing the right things. “You don’t have to outrun the bear,” he says of his strategy. “You just have to outrun the other guy.” Read NPR’s full profile of Craig at the source link below.
Follow topics and authors from this story to see more like this in your personalized homepage feed and to receive email updates.
