The structure and static topology of social networks has been the subject of multiple network theory studies. In game theory, the prisoner’s dilemma, which strongly outlines the difference between the cooperation present in communities and the selfish objectives of the individual, has also received much interest.
In my research on a sample of 233 people, described in my thesis, I examined how a real social network behaves in an iterated prisoner’s dilemma situation where the participants may decide which other partners they are willing to play with. My assumption is that this model strongly resembles the cooperation of real-life communities.
The results show that cooperation is much more stable in real life communities than networks of simulated agents. The predominant majority of people consistently play a selfless cooperative strategy. When faced with competitive opponents, most of players respond by breaking off contact. Length of cooperative partnerships follows a long-tail distribution: most of partnerships are very short, but there are also examples of very strong friendships. If the players are not aware of each other’s payoffs, and are forced to choose partners based only on trust, the inequalities of profits decrease radically compared to real life examples.
Besides the findings of my experiment, I describe the design and development of the web-based software used, the technologies chosen for the project, the main obstacles throughout the development and the solutions to these problems.