Four signs America's broadband policy is failing

In 2008, I wrote a paper for the Cato Institute questioning the need for network neutrality regulations; I argued that the Internet's decentralized architecture made it inherently resistant to mischief by broadband incumbents. While I'm still skeptical about the wisdom of network neutrality regulations, I've become more concerned about the state of the broadband market in the four years since writing that paper. In a March article for National Affairs, I made a case for regulatory action to prevent further consolidation of the largest broadband firms.

What changed my thinking was less the theoretical arguments set out in that piece than it was a sequence of developments in the telecom marketplace. It forced me to reexamine my own assumptions about the state of the broadband market. Here are the four most important.

elecom policy wonks have held a long-running debate about how the United States stacks up against other nations when it comes to Internet access. In 2009, a team led by Yochai Benkler at Harvard's Berkman Center produced a voluminous report on the subject which found that broadband service in the United States was distinctly mediocre.

The report attracted a harsh response from some libertarians, including Brett Swanson. What was striking about Swanson's response (and others were similar) was that it didn't seriously dispute the core findings of the report. For example, Benkler's team found that the United States had gone from leading the world in broadband penetration to being ranked 15th by an OECD report. Swanson countered that if you crunch the numbers better, the US is actually around eighth or tenth—which is to say, also not near the top.

That matters because a key argument for America's relatively hands-off approach to broadband regulation has been that giving incumbents free rein would give them incentive to invest more in their networks. The United States is practically the only country to pursue this policy, so if the incentive argument was right, its advocates should have been able to point to statistics showing we're doing much better than the rest of the world. Instead, the argument has been over just how close to the middle of the pack we are.

One reason I was sympathetic to Swanson's argument in 2009 was that Verizon was then in the middle of a massive investment in its new fiber optic network called FiOS. AT&T was also replacing parts of its copper network with fiber. I expected this would spark an arms race between the telephone and cable companies, leading to rapidly increasing speeds.

Instead, in 2010 Verizon announced that it would stop installing fiber without reaching some of its most important markets, including Baltimore, downtown Boston, and my own apartment in Philadelphia. It now appears that none of the "Baby Bells" have any further plans to run fiber optic cables to peoples' homes. That means only the minority of households with FiOS service (and perhaps some of AT&T's U-Verse customers) have an alternative to their local cable company for faster-than-DSL connectivity.

Verizon underscored its dwindling interest in the wired broadband market earlier this year when its wireless subsidiary signed a deal to sell Comcast broadband service in Verizon Wireless stores. Instead of an arms race between telephone and cable incumbents, we seem to be getting a truce.

In my 2008 paper, I made an in-depth argument for the long-term stability of the Internet's decentralized architecture. I won't rehearse the explanation here, but a key premise was that the Internet consisted of relatively small networks at the edges of the Internet that paid other, larger networks for access to the Internet's backbone. In the center of the network were several "tier one" backbone providers who peered with one another on a settlement-free basis and did not have to pay anyone else for connectivity. Competition among these tier one providers ensured that no single firm wielded too much power.

So I was surprised when Comcast forced Level 3 to pay it for connectivity in 2010. (Level 3 had become the content delivery network for Netflix.) Until then, Comcast had been paying Level 3, a "tier one" backbone provider, for connectivity. Given that I'd written a whole paper based on the assumption that payments flow from edge networks to backbone providers, I felt a bit like a geologist who suddenly spotted water running uphill.

There's still significant dispute about what happened and whether Comcast did anything unethical or illegal. But the incident is a clear sign of Comcast's growing bargaining power relative to other major networking firms. And that's cause for concern because, while there are plenty of alternatives to Level 3's transit services, only Comcast can deliver traffic to Comcast's 17 million broadband subscribers.