We now assume that we have an accurate model of the state, and we've received some sensory input from the simulator for the next time step. We have to update our model to account for this input. We divide this update into two stages depending on what caused the changes to the model. First, we update the model for internal changes, those that are caused by the client (e.g. turning and dashing have a large impact on the model since the positions that are stored are relative to the client.) Second, we update the model for external changes, those that are not directly caused by the client (e.g. the motion of the ball and other clients.)