The joy of Albert-László Barabási's book, Linked: The New Science of
Networks is that, after reading it, you can't look anywhere without
seeing networks. The book, which is currently ranked 99th on Amazon's top
seller's list, leaves a powerful imprint on the mind's eye.
Did you realize that it takes an average of only one link per node to
bind together a random conglomeration of 100 nodes into a seamless network?
No wonder gossip travels so fast.
But although most laymen and scientists imagine a "classic network" to be a
random and evenly distributed mesh of linkages among nodes, Barabási
illustrates that many key networks are, in fact, severely uneven. In these
networks, called "scale-free," most nodes have only a few links, while a few
nodes have lots.
Obviously, to anyone who looks at their own social network, the scale-free
model doesn't seem radical. We take for granted that our neighbor Howard is
in close touch with 100s of people while most people, like his wife Susie,
talk to the same 35 people year in and year out.
What astonishes, however, is that nearly every network we know — whether
it is the network of a cell's molecules, Hollywood, the Internet,
Yellowstone's ecosystem, or the "sex map" of patients suffering AIDS — is
populated almost entirely by Franks and Susies. None of these networks
conform to the classically imagined "random" model. All these are
free-scale networks. All rely on a few hyperlinkers to do the bulk of their
work, whether that work is communicating, making movies, eating rodents, or
passing on disease. Moreover, these uneven networks display distinct and
recurring distributions of high-linkage versus low-linkage nodes.
The scale-free network's linkage distribution has important byproducts. Like
the "classic" random network, scale-free networks are amazingly efficient at
distributing information. Unlike their random cousins, scale-free networks
are practically immune to the random failure of individual nodes — nearly
all nodes can be eliminated and the network survives.
Article comments