There’s a scene in the Steven Soderbergh movie, Traffic, where the widow of a drug dealer brings a doll to the Columbian drug kingpin. “The doll is stuffed with cocaine. Big deal, we’ve been doing that for years,” he says dismissively. “No,” she answers, “the doll is cocaine.” The whole toy is a heat-treated, compression-molded block of cocaine, undetectable to sniffing dogs. The drug lord becomes very interested.

The Internet is like that doll… and not because it’s used by some for smuggling drugs! Rather, the Internet is seen as a thing filled with interconnection relationships, when in fact the Internet is interconnection. The relationships make the internetwork. They are more significant than the TCP/IP protocol, the end-to-end design philosophy, the bandwidth, or the routing algorithms, as important as all those things are. Kill interconnection, and a network disappears from the Internet. Kill the culture of interconnection, and the Internet dies. Another analogy is Arthur Koestler’s concept of a holon—something that is both a whole and a part of the whole. (Thanks to Miko Matsumura for the pointer at a recent retreat.)

The value of interconnection is often missed, because it’s the space between networks. It’s much easier to grasp the impacts of those individual networks on their customers. Every piece of the Internet, however, must interconnect to serve its users, which means its internal policies and practices are never the whole story. Interconnection is generally reciprocal, so if you want to benefit from a link with a network, you take on some obligations in return. The details get complicated, and network interconnection is constantly evolving, but that’s the core magic.

Today, John Markoff published a New York Times article on how Internet interconnection may be changing. (The short version is that private peering is short-circuiting the major backbones, with unpredictable consequences.) Markoff deserves credit for giving a serious summary of academic network science research that bears on Internet structure. You usually don’t see these concepts in the popular press. It matters whether or not the Internet is a scale-free network, however, as esoteric as that may sound. As Markoff notes, even the experts can’t agree on what the Internet looks like today, raising serious questions about its performance going forward. They just know that it’s changing. One reason is the lack of public traffic data on Internet-connected networks, which KC Claffy of CAIDA has been warning about for years.

I wrote three law review articles about interconnection over the past three years. I didn’t realize it, but they form a trilogy. Only Connect argues that interconnection, not non-discrimination, should be the central focus of telecommunications policy today. The Centripetal Network delves into the network science that Markoff’s article summarizes, raising the concern that the Internet’s interconnectivity may not be as robust as it seems. And in Off the Hook, coming out shortly, I develop a detailed legal theory for an interconnection-based policy regime under the Communications Act.

Interconnection is poised to become even more important, because it’s not just a factor at the network layer. Internet applications and content are increasingly becoming interconnected, moving toward the syndication model of business I proposed a decade ago. Twitter interconnects with Google for real-time search, while YouTube interconnects with blogs for content distribution. Everyone’s a platform, and virtually everyone is both a consumer and a producer of external information. I’m firmly convinced that the dynamics of interconnection will keep policy-makers and business executives busy for years to come. All the more reason to make it a focal point now.