Correlated decision making: a complete theory — LessWrong