#interactiveclass

However, with games and with lectures in the social sciences, there is often some detachment of what is taught in classrooms to reality. For example, the classic Prisoner’s Dilemma is often used to illustrate the concept of Nash Equilibrium. Yet, students often bring up the fact that two criminals may commit many crimes together over time and they might cooperate (not rat each other out) knowing if they rat out their partner now, they may not be so lucky next time. Simulating repeated interactions like that in class, when the time is limited and where the number of students is unpredictable, can be especially difficult and may just be outright impossible with a paper and pencil based setup. To make a repeated game work in the classroom and deliver the expected learning outcomes, the following components are necessary:

Maintain Matching

Without being able to match the same players together during the course of a repeated game, it would be difficult for students to test any strategies, including the ones they developed themselves.

Provide Instant End-of-Game Feedback

Students need to know the outcome from each game so that they can learn and adjust their strategy, if necessary.

Simulate Infinitely Repeated Interactions

If students know the number of times the game will be played beforehand, they may be able to ‘solve’ the game through backwards induction. Although this can be an interesting lesson, it is different from simulating infinitely repeated games. The ability to create that uncertainty of when an interaction may end is a must for repeated games to emulate the real world.

Deliver Aggregated Results

One of most of interesting aspects of repeated games is to allow students to spot trends, if any, during the course of the game. Being able to show results by looking at the repeated games as a whole, i.e. as a super game, offers a powerful way for students to see the emergence of certain strategies and evolution of the relationships, immediately playing it themselves. MobLab has made running repeated games almost effortless. Built on top of the existing ‘Replay’ function, where an instructor can repeat a finished once at a time, an advanced set of parameters are now added to streamline and automate both finitely and infinitely repeated games. Here are some of the highlights:

We have implemented a new parameter for all games called Minimum Periods. This is a threshold where the repeated interactions would last at least the specified number of periods.

To simulate infinitely repeated interactions, instructors can set Ending Probability, so that after the Minimum Periods has been reached, this is the probability that the game will end following the next period. This creates the uncertainty that is parallel to real-world applications.

In addition to the existing per-game summary results, a few of our games have multi-period game results and aggregated statistics to track how each group progressed in the multi-period game as well as how they compare to other groups.