opening it up with Common Lisp
Book review: Darwinia
Summer reading: Spin
the Omnivoire's Delimma
the Golem's Eye
Newman and Juyong investigate why social networks differ from most others in that the degrees of adjacent vertexes in social networks tend to be positively correlated whereas the they are negatively correlated in others. Why is it, in other words, that people with lots of friends tend to have friends with lots of friends but that sites with lots of links tend to link to sites with fewer links?
In a beautiful bit of mathematics that, sadly, is mostly beyond my current ken, they show that simply adding organization (grouping) to a network creates a strong tendency towards the positive correlation. This grouping also can explain why social networks have higher clustering coefficients (i.e., it's likely that I am a friend of my friend's friend).
They conclude their paper with two investigations: one of paper collaborations among physicists and one of links between members of boards of directors. In the first case, grouping alone appears to account for all of the correlation whereas boards of the directors have an even higher correlation than the grouping can explain. In other words, the particularly high correlation among directors appears to have an actual social component.
Now I need to go back and re-read the paper with an eye towards actually understanding the math!
Copyright -- Gary Warren King, 2004 - 2006