• by Noah Brier , Paul Woolmington , August 2, 2007

From Facebook to Dove's "Real Beauty" campaign, social networks play a key role in media today. But, despite the current hype, networks aren't particularly new. Consider, the famous "six degrees of separation" experiment examining social networks was conducted in the 1960s.

What's new is our understanding of how they really function. Just 20 years ago, if you asked a social scientist to graph out a social network they would have shown you a bell curve: Most people are friends with a few people who know hardly anyone; a few who know a massive amount of people; and the majority sit in the middle. Bell curves, or Gaussian distributions, traditionally result when the subjects being plotted aren't connected to one another.

But when scientists actually looked at how links were distributed on the Web, they didn't get a bell curve at all: There was no fat middle full of sites with some average number of incoming links. On the contrary, there were a few sites like Google, Yahoo, Microsoft and eBay that had massive amounts of incoming links; the rest of the world was insignificant in comparison. In other words, the graph looked like a hockey stick on its side, not a bell.

With this discovery, scientists went back and re-examined those networks they thought they understood. The more they dug, the more bells were replaced with hockey sticks. Even social networks, it turns out, are radically unequal. We all know a few people who have a massive number of contacts and the rest of our friends are completely disconnected in comparison.

I wish I had come up with all this myself. However, the credit goes to Albert-Laszlo Barbasi, who wrote about much of it in his 2003 book Linked: How Everything Is Connected to Everything Else and What It Means. "Why did we have to wait until 1999 to discover the impact of hubs and power laws on the behavior of complex networks?" Barbasi writes. "The answer is simple: We lacked a map. The few network maps available for study before the late 1990s had a few hundred nodes at most. The enormous World Wide Web offered the first chance to examine the intricate anatomy of large complex systems, and established the presence of power laws. As other large maps followed, we gradually understood that most networks of practical interest ... are shaped by the same universal laws and therefore share the same hub-dominated architecture."

Chris Anderson's "long tail" is another good example of power laws at work. Anderson pointed out that when you look at how commerce, and most everything else, works online, you see a power curve. What it illustrates is radical inequality: A few "hits," as Anderson likes to call them, dominate and everything else is a "miss."

We've only begun to dig into what all this means for the future of media. Of course, the long tail starts to hint at the rise of niche media. But there are some other fundamental characteristics of networks that may help us understand what's coming next. Take, for example, the ubiquity of power curves:

When you look at the linking patterns of the whole Web, a few huge Web sites emerge and the rest are dwarfed in comparison; but this pattern also holds for the smaller subgroups. For instance, when you just look at marketing blogs, you see the exact same patterns. A few players dominate the incoming links, while the rest exist within a very small universe of connections. No matter how niche you get, this behavior stays consistent.

In fact, I think many have misinterpreted Anderson's long-tail thesis. The tail, it turns out is not flat at all. Rather, if you zoom in on any section of it - that is, if you plot influential sites in any particular niche - you'll notice the exact same hockey stick on its side you see when you pull back and look at the whole Web. This phenomenon is important, because it shows you where the opportunities are in each specific subgroup, regardless of size.