Blackbox is a computational thinking game presented as part of a class facilitation at Cornell, on the topic of transparency and algorithms.
One player acts as a binary classification algorithm given some procedure (ACCEPT or REJECT given an input), and the other players have to figure out what the procedure is. Players get colored cards to feed into the algorithm, while the player serving as the algorithm gets a random card describing what algorithm to follow.
The full structure of the game is as follows:
- Each group has 4-5 members.
- One person volunteers to be the algorithm, the rest are users.
- The users are given a deck of colored cards with four colors, RED, BLUE, YELLOW, GREEN.
- The algorithm only accepts one input card at a time and decides whether it is accepted or rejected.
- The objective of the game is for the users to figure out the algorithm.
This game could be used in an educational context to teach how algorithms work and talk about ethical issues surrounding algorithmic accountability and transparency.