‘Collaborative networks’ – such as the pattern of links between scientists who have worked together – are good candidates for study because they tend to be clear-cut: the links between people are well defined, the dates of the associations are clear, and the details are often logged.

Alberich’s team realised that the relationships between the characters in Marvel comics – there are dozens of titles in total – formed an artificial collaborative network. Characters from one comic frequently appear in another, and Alberich and colleagues viewed a ‘collaboration’ as each time two characters turn up together. They analysed around 96 000 appearances by 6500 characters in 13 000 issues. The data were gathered by the Marvel Chronology Project.

Most real collaborative networks are ‘scale-free’, that is the number of people with links to others falls as the number of links grows. Alberich’s team found that the ‘Marvel Universe’ was also scale-free – most characters have appeared with an ‘average’ number of other characters, but fewer are associated with many. This relationship has a cut-off point: with 1625 appearances, Spider-Man is the most ‘connected’ character.

“Every fan believes that the Marvel Universe is a real place”, team member Francesc Rossello told PhysicsWeb. “Now we have shown that this perception has a mathematical basis”.

But the Marvel Universe lacks the ‘clustering’ effect found in real networks, in which two people are more likely to be connected if they are both linked to a third individual. Alberich and co-workers attribute this to the writers, who had to distribute super-heroes evenly among the comics.

The investigation – which was funded by the Spanish government – concludes that although the Marvel Universe successfully mimics many aspects of human networks, it cannot disguise its artificial origins. Alberich and colleagues now plan to study the evolution of the Marvel Universe, in order to establish the factors that lead to differences between social networks and completely random networks.

Alberich’s team has submitted its study to the journal Social Networks.