Applying the Prisoner’s Dilemma to Quantify Community
While looking for statistics on suspect confessionals during arrests (in an attempt to discern a typical response to the prisoner’s dilemma), I came across this article by Max Nisen, which examines the outcomes of an experiment conducted by two researchers from the University of Hamburg. The experiment applies the prisoner’s dilemma to actual prisoners (and is allegedly the first of its kind), incentivizing participants with coffee or cigarettes. The researchers conducted two versions in the game: one in which participants had only one chance to play (the simultaneous version) and one in which participants would play multiple times in succession (the sequential version). What made the study especially interesting to me is that the researchers repeated the study using students as participants.
The researchers found that in the simultaneous version of the game, the prisoners picked cooperative choices 56% of the time, while students only cooperated 37% of the time. In terms of obtaining the best “mutual outcome” (maximum net gain), students were successful 13% of the time while prisoners were successful 30% of the time. Nisen suggests that these statistics imply that prisoners (or at least this particular set of prisoners) are more cooperative than students (or, again, this specific group of students).
Nisen introduces a unique application of the prisoner’s dilemma: quantifying trust within a community. While the experiment has its flaws, it can potentially be used to gauge a network’s willingness to cooperate. We can also use the experiment to test trust within a pair of people – picking mutually cooperative, non-greedy choices could hint at more “care” within a relationship. Furthermore, the experiment can be expanded to test how trust changes when the size of a network increases – my hypothesis would be that there is greater trust in smaller networks, as smaller networks would likely have a higher frequency of strong ties.
Though the experiment can possibly be implemented to obtain a quantitative measure of “community,” it does have its flaws. Firstly, I was unable to find the age of the students participating in the study or any statistics on the amount of time they have known each other. While the prisoners were all prisoners, it is not explicitly stated that the students were similarly homogeneously female. Varying the age and gender of the students could drastically impact the results of the experiment. For example, younger students may be more inclined to pick greedy, non-cooperative choices or may be less likely to cooperate with students of the opposite sex. Furthermore, the “net gain” for each system is similarly different – cigarettes and coffee likely hold more value to someone living in a prison cell than euros hold to students who can afford a college education. Yet despite these imperfections, modifications of the experiment (which would require scaling “gains” to the population of interest) can still lead to meaningful insights on connectivity and cooperation within a network.