Today’s Oregonian has an interesting little article on the world of game programming centered around Orego, a computer program in development at the Math Department of Lewis & Clark College that plays Go.
I remember Go-playing being cited as one of the challenges in pattern recognition and AI development back when I was first getting into programming, it’s interesting to see how people have approached the problem over the past thirty years.
The article mentions that professor Peter Drake and his team had “tried nearly every novel approach out there: neural networks, cellular automata, and genetic algorithms” over the past few years. What they’re experimenting with now is known as a Monte Carlo method. Instead of using a specific strategy or algorithm to solve the problem, it iterates a series of random moves until a solution is reached, then does it again and again, then chooses which combination gives it the best results.
Sort of reminds me of the brute force approach to the proof of the four color theorem. For a guy who’s used the motto “Multimedia Design by Brute Force” for more than fifteen years, you know that has some appeal.
BONUS: The accompanying “Factbox” on Go notes that “The game reportedly has been played by such luminaries as Albert Einstein, Bill Gates and Rod Stewart.” Some company, that.