A Nondiscriminatory Net: The Right Approach

This is the first in a series of posts by Chris Riley, Free Press Policy Counsel, to summarize the primary policy recommendations made in recent comments submitted to the Federal Communications Commission in its open Internet proceeding. Today’s topic: nondiscrimination.

The FCC is in the process of creating a Net Neutrality rule, and front and center in the proceeding is the proposed rule of nondiscrimination – the idea that network gatekeepers should not abuse their control over the pipe by discriminating in favor of or against any online communications. The first of the existing FCC principles defends the right of Internet users to have access to content, applications and services – but it says nothing about the quality of that access.

Under the existing principles, network operators like AT&T and Comcast could speed up their own content, applications and services, and could slow down their competitors’ – an especially troubling proposition when the nation’s largest fixed Internet access service provider, Comcast, announces a merger with one of the nation’s largest content companies, NBC. Network gatekeepers can also use discrimination to impede services like Voice over Internet Protocol and online video – slowing down their bits can render them undesirable to users – and can maintain the stranglehold that the gatekeeper currently has over voice and video services.

But the harms here are not to competition only. Network gatekeepers can also reduce or delay investment if permitted to degrade any service that isn’t popular enough, steadily turning the Internet into a walled garden and stifling innovation and evolution in the process. They can also increase their already padded revenue streams – while reducing investment in their networks – if they’re allowed to charge exorbitant fees to content, applications and services companies for the privilege of using their terminating access monopoly. Sounds like a great proposition for Wall Street, but it would be the end of the Internet as we know it.

To stave off these dire harms, the FCC has proposed a fifth principle and a rule on nondiscrimination intended to prevent network operators from discriminating against or in favor of different uses of the Internet. The language proposed in the agency’s NPRM is clear: “Subject to reasonable network management, a provider of broadband Internet access service must treat lawful content, applications, and services in a nondiscriminatory manner.”

Defining Nondiscrimination

Last week, Free Press filed comments with the FCC in support of a strong Net Neutrality rule. Our filing supported the FCC’s language – with a big caveat that the terms “nondiscriminatory manner” and “reasonable network management” must be defined correctly. “Reasonable network management” is a tricky issue, and I’ll tackle that in a later post. Today’s focus is on “nondiscriminatory” and the different options on the table for that concept. The summary version of our position is that a clear and complete rule that prohibits all discriminatory behavior is essential to avoid loopholes; provide clarity for consumers, network gatekeepers, investors and Internet developers; and protect consumer choice, competition and innovation.

Our comments focus on the two most prevalent definitions for a nondiscrimination principle: a clear rule against discrimination subject to reasonable network management, versus a rule that instead prohibits only “unjust and unreasonable discrimination.” Sometimes the second option uses other language, such as AT&T’s proposed “unreasonable and anticompetitive discrimination” standard, but regardless of the qualifiers, the proposal is the same: Declare some forms of discrimination to be beneficial, without looking at the network management practice itself.

The first big problem with the “unreasonable discrimination” approach is that it’s utterly unworkable. The goal of powerful network gatekeepers in this proceeding (and, well, every other regulatory proceeding) is to create rules that are made of mush, with giant loopholes that their armies of highly paid K St. lawyers and lobbyists can get through. So their ideal fifth principle is as vague and as devoid of limitations or meaningful standards as possible. That serves their purposes admirably.

What does “unreasonable discrimination” even mean? Is it “reasonable” for a network operator to block BitTorrent or another peer-to-peer program, just because it sometimes uses a lot of bandwidth? That’s in the eye of the beholder. It might seem to network operators that selectively targeting that application makes sense. But for consumers, they’re paying for that bandwidth, and how they use it should be up to them; no network gatekeeper should be allowed to interfere with their lawful use of the Internet, so any discrimination should be unreasonable. And a rule that prohibits only “unreasonable discrimination” is a rule that leaves unfettered discretion to the FCC to choose between consumers and network gatekeepers in each individual case, providing no certainty for any of the affected parties (and no clear standards on which a reviewing court could hang a reversal).

Application Bias

We spend a section of our comments on one particularly insidious idea: application bias. The idea is that a network gatekeeper declares some applications to be “high priority” and others “low priority,” and engineers the network to favor the “high priority” applications. One problem with this over-engineering is that it’s unnecessary. The Internet has performed well without it when network gatekeepers respond to growth in usage by adding capacity. But the biggest problem is that the network gatekeeper is a terrible proxy for the user in determining what is, and is not, high priority. As we showed in a white paper, The Hidden Harms of Application Bias, these mistakes can and will harm innovation on the Internet, and will create advantages for incumbents that undermine user choice and competition.

Here’s the bottom line: Any form of discrimination harms some traffic. It slows down some data traveling on the Internet, and speeds up other data. Yes, it’s possible for some forms of discrimination, in some contexts, using some techniques, to create a net positive for the Internet. But there are a lot of qualifiers there, and all of them are important. Any rule that makes a blanket exception to declare some forms of “discrimination” to be acceptable, without any contextual analysis, would harm consumers, competition and innovation. The right approach is not to create a mushy, vague rule with built-in K St. loopholes, but to prohibit all discrimination, and then to evaluate potentially beneficial discrimination through a clear (and standalone) framework of reasonable network management.