1: \begin{abstract}

2: This article investigates selfish behavior in games where players

3: are embedded in a social context. A framework is presented which

4: allows us to measure the \emph{Windfall of Friendship}, i.e., how

5: much players benefit (compared to purely selfish environments) if

6: they care about the welfare of their friends in the social network

7: graph. As a case study, a virus inoculation game is examined. We

8: analyze the corresponding Nash equilibria and show that the Windfall

9: of Friendship can never be negative. However, we find that if the

10: valuation of a friend is independent of the total number of friends,

11: the social welfare may not increase monotonically with the extent to

12: which players care for each other; intriguingly, in the

13: corresponding scenario where the relative importance of a friend

14: declines, the Windfall is monotonic again. This article also studies

15: convergence of best-response sequences. It turns out that in social

16: networks, convergence times are typically higher and hence

17: constitute a price of friendship. While such phenomena may be known

18: on an anecdotal level, our framework allows us to quantify these

19: effects analytically. Our formal insights on the worst case

20: equilibria are complemented by simulations shedding light onto the

21: structure of other equilibria.

22: \end{abstract}