Context: Football
In the football scenario, the Agent is either a professional football player or a pre-programmed football-AI. This agent then, during a point in the game, is faced with either passing the ball, or shooting to try and score. Either way, this can result in a goal being scored off of the play, or not.
The end outcome is then either a win or a loss at the end of the game, and that could be by various margins. For the sake of this experiment, assume that you support the team for which the agent is playing.
The question is: to what extent did the football agent cause their team to win/lose?
An example of this is given below:
| {{ Button1 }} | {{ Button2 }} | |||
| {{ Button3 }} | Next button (usually) | {{ Button4 }} | ||
| {{ Button5 }} | {{ Button6 }} | |||
| {{ Button7 }} |
From the above, you should see that the human agent, who receives bonuses based on how many goals they score, was playing at home ground, tried to shoot for goal and missed, but the team still won the match by 3 points.
Note that each bolded section of the above summary corresponds to one interchangeable information block. Note that how much you feel the Agent caused the Outcome (win) might depend on any number of the above points of information.