Game theory is an interesting topic with a variety of uses due to its subject matter; strategic decision making. It tries to model how rational intelligences will work together given certain rules and goals. In other words, it predicts what the players of any system with a game-like structure will do. Researchers at the University of Manchester have recently suggested that the current basis of many of its models may be flawed though.
A large part of any game theory model is information as it is a player's knowledge of the game and players that will determine their actions. Typically the models are based on the equilibrium point, but this assumes the players are making the best decisions for themselves, while taking into account what other players are doing. The researchers are stating that this is not possible for complex games such as chess, Go, and even the stock market to be modeled with the equilibrium assumption because the complexity is too much for a human to understand.
Thus far the researchers' results indicate that increasing the player count for a game decreases the system's ability to reach equilibrium. Hopefully with more time they will be able to determine a new means of predicting player behavior, and thus create better models for some systems.