The new science of networks

When the Internet and the World Wide Web ballooned in size during the 1990s, they provided the first very large networks that could be measured and analyzed via crawlers and worms in the same manner as the search engines do their job. Meanwhile, social networks-networks of relationships among people-became accessible for study via large databases that documented collaboration in fields ranging from scientific publishing to movies. As the measurements started to come in of the distribution of router node sizes, website-linking statistics or social connections in Hollywood, the findings were totally unlike the predictions of random network theory in two fundamental ways. Small-World Networks First, real-world networks tend to be clustered rather than being a uniform distribution of links. Typically, they're small, densely interlinked clusters of nodes with relatively few links to other clusters. The second surprising attribute of real-world networks is a non-normal distribution of node sizes.