Let’s Play Rock-Paper-Scissors
Against the computer that is. Or at least, a computer algorithm.
A recent article on the New York Times features another interesting example of the human vs artificial intelligence, this time over a game of rock-paper-scissors. The flash applet embedded in the article allows the you to play rock-paper-scissors by clicking the particular move on the “Human” side. At the same time, the “Computer” side also makes a move, unaware of the choice made by the human player. The article states that in game of rock-paper-scissors completely left to chance, the tally of wins, losses, and ties would be split evenly from the total number of games played. However, the “computer” operates on an algorithm that uses past choice patterns of the player to predict future moves. The algorithm takes the last four moves made by the player and searches the list of past moves for the same pattern. If an identical set of four moves is found, the algorithm makes a prediction about the possible next choice the player would make. If no identical set was found, the algorithm would then search for a smaller set of moves (e.g. set of last three moves made by the user) until it can make a prediction. Since the game does not give the option of seeing the computer’s thought process until the player has made a least five moves, it is safe to assume that the first few moves is completely random.
Does this actually work? Thinking back to the last time I played this simple game, I don’t think every choice of rock, paper, or scissor I made was random, there was definitely a good bit of predicting involved in an attempt to outsmart my opponent. The algorithm doesn’t take this into account, that the choice made by the human player is in response to it own moves. I gave the game a try and, with the knowledge that the computer is making predictions based on my past moves, deliberately switched around with my choices. The score came out to be 58-42-50.
Go ahead and try it for yourself!