Good analysis of how the social networks of Marvel comics characters differ from real social networks.
Alberich's team has studied the statistical properties of the network of 6,486 characters in the 12,942 Marvel comic books. On average, each book features just over seven characters; one features 111.
The probability of a book containing a certain number of specific characters depends on the group size, the team found, at least for groups of ten or more. To this extent the Marvel Universe resembles real networks.
A closer look reveals the Marvel Universe's artificiality. For example, social networks have a property called clustering: two people who share a common friend are more likely to know one another than are two people chosen at random.
The Marvel network is only very weakly clustered – about 1.5 times more than a random network. Clustering in real networks is typically ten (or more) times greater than in random webs.
(Thanks, James!)
