Contagion,
apart from being a recent movie title, is something we all shy away from,
especially when it is attached to ’flu via the name of an animal such as a pig
or a chicken. The usual idea is that it (the contagion) will spread faster, the
more contacts that people have who are harboring it.
The
assumption that it is best to avoid contact is a sound one, but other factors
come into play as every contact is not the same as every other one. The
modeling of contagion has become a fertile ground and it is easy to see why
this is important in the prediction of a disease pandemic.
But
it is not only disease that can spread like wildfire through a population.
Fashions do too, whether they are based on style, the latest gadget and now
being a member of the right social network.
All
this adds to the pressure to understand the contagion model that is going to be
a robust predictor of the spread via contact. Ugander et al in this week’s Proceedings of the National Academy of Science
have data-mined the registration of new Facebook users for the year 2010 (1).
Anonymity was maintained as the users were just nodes in the network, so any
one of us was just one of almost a billion users on the day chosen that year.
The
focus was sharpened to 10M users with a variety of friend numbers after a week
and an interesting result emerged. The numbers of contacts wasn’t the driver
for being socially contagious, rather the diversity of the contacts was more
important.
This
has some interesting implications as it seems that we like to connect with
people coming from diverse backgrounds, points of view etc., although the usual
concept was that we tended to stick in our own type of group. It seems that the
Internet is a bigger force for democratization than we’d realized.
