Over the weekend,
The New Yorker called Comcast a “menace” and a “racket” and denounced the deal as “a tax on the larger economy.” Not to be outdone, the Verge declared: “The Internet Is F-----.”
Despite America’s global lead in apps, content, and broadband services (both wired and mobile), advocates of more Internet regulation continue to promote the idea of U.S. Internet decline. The zealous insistence that Web firms like Netflix route their traffic inefficiently is similarly bizarre and can only be explained by ideology or an information deficit. Ideology is difficult to overcome. A better understanding of how the Internet works, however, may soften these eruptions and also help Washington navigate a proliferating number of technology policy matters.
From the beginning, smaller networks and content providers always paid other networks for access. Consumers, businesses, and websites paid Internet service providers (ISPs), who paid regional “Tier 2” ISPs, who paid even larger continental or global “Tier 1” ISPs. Networks that exchanged roughly equal amounts of traffic, meanwhile, often engaged in no-cost “settlement free peering.”
With Moore’s law advances in hardware and corresponding upheavals in software and content, the Net’s architecture and business models have always been in flux. Big new technologies — fiber optics, the World Wide Web, broadband access networks, Web video, mobile, Wi-Fi, the cloud — change the technical and economic realities of the network. And so we build yet more networks, data centers, and software stacks — over $1.2 trillion worth in the U.S. since the late 1990s.
In a new report called “How the Net Works,” coincidentally released just before the Comcast-Netflix news broke, we offer a brief history of these interconnection practices: how networks connect; who connects to whom; on what terms; how network topologies and business models shifted over time; how content delivery networks (CDNs) and the “Hyper Giants” like
Top-down direction, however, is what the Net pessimists want. Columbia law professor Tim Wu, a key backer of “net neutrality” and coiner of the term, for example, insisted on his New Yorker blog that “This is the first-ever direct interconnection deal between a broadband provider, like Comcast, and a content company, like Netflix or Google.” The intention is to scare, implying something drastic has changed and the Internet is at risk. But the statement is flat out false.
Many agreements just like Comcast-Netflix already exist. Firms like Google, Amazon, and Microsoft have for years connected directly to broadband networks via paid peering arrangements. They want to make sure their products — like search, email, ecommerce, video, and cloud applications — deliver the fastest, most robust bits to consumers and businesses. Their choice? Pay a network company for an indirect route to their customers. Or pay a network company for a direct route to their customers. Sometimes, for some services, the lower-cost indirect route makes perfect sense. Other times speed is crucial. CDNs like Akamai and Limelight, which store third-party content closer to consumers, have long employed the direct-connect paid peering model.
The alarmists think this is all about big network bullies exploiting helpless content firms. They are demanding, just as we predicted, that net neutrality rules, which didn't cover interconnection and peering and which courts have thrown out twice, now be extended to the entire Net. But there is little evidence big firms are crushing the little guys. After their hook up was announced, Netflix stock jumped, and Comcast declined. Or consider mighty Comcast has 3,250 times the revenue of tiny WhatsApp ($65 billion versus $20 million) but just seven times the “market cap” ($132 billion versus $19 billion). WhatsApp acquirer Facebook, of course, is worth $178 billion, a third more than Comcast, and this week the private Web document sync firm Dropbox was valued at $10 billion. So tell us, just who are the big guys?
The Internet is a success of self-organization and self-governance. Building something so complex requires exquisite planning by individuals and teams creating the hardware and software to power such a sprawling system. It also requires a conceptual framework that provides just enough commonality to make the pieces work together, but not so much top-down instruction that the system cannot adapt, grow, and innovate. Thank goodness we didn’t, over the last 40 years, listen to any individual or central committee who presumed to know exactly what network architectures and business models would prevail every step of the way.
