Data traffic prioritization is one of the most unfairly maligned technologies. Caricaturing commonplace network management techniques as “fast lanes,” net neutrality activists warn that introducing the option of paying for specific performance levels of Internet traffic will destroy the characteristic “openness” of the web. This is false. Prioritization and other mechanisms that differentiate data traffic are the only economical ways to radically improve the performance of broadband given the wide diversity of different applications that broadband networks must support, and would encourage further innovation throughout the Internet.
When it comes to discussions of potential net neutrality legislation, there is fairly widespread agreement on what the substance of rules should look like. There is rough consensus around blocking and throttling—namely, that baseline rules should flatly ban Internet Service Providers (ISPs) from blocking legal Internet traffic or slowing or throttling data to extract payment. But the rough consensus around substantive rules admittedly breaks down at paid prioritization.
Prioritization or other forms of traffic differentiation can optimize Internet Protocol–based (IP-based) networks for different purposes. Indeed, from the Internet’s very beginning, engineers have designed protocols for traffic differentiation precisely because the Internet is a best-effort network, in which not all applications work optimally without differentiation. Unlike circuit-switched telephony, where there was one wire and one application, the Internet supports a dizzying array of applications, many of which work just fine over a traditional neutral network, while others would be greatly improved by prioritization.
Prioritization is not a zero-sum proposition—that is, it is possible to see a benefit without an offsetting loss. In some cases, this is because slight delays in packet delivery are far less noticeable to consumers for some applications than others. For example, nobody would notice email packets being delayed by 50 milliseconds, yet such a delay would make a big difference in video conferencing applications. Prioritization works by optimizing the timing or scheduling of packets for different applications, allowing tradeoffs to not detrimentally affect the quality perceived by a user. It is possible to send email, stream videos, and build new tools for communication, continuing the characteristic openness of the Internet that has generated such tremendous innovation in recent decades, all while allowing for traffic differentiation that enables real-time applications with very strict performance requirements, especially related to the delay or variability in delay of the data flow.
Most all agree there should be room for “specialized services” that run over the same infrastructure as the Internet, such as for telehealth applications. The question is how strict or permissive restrictions should be on traffic differentiation. A relatively permissive regulatory regime would allow for all companies—large and small—to contract for prioritization services, and not have to navigate a bureaucratic process at the FCC or arrange for specialized billing with broadband providers.
In terms of potential net neutrality legislation, Congress has the opportunity to craft very fine-tuned institutional arrangements and specific oversight processes. It is important this be done in a way that provides lasting certainty and clarity for all actors in the Internet ecosystem, and also preserves the continued evolution of networks to support increasingly important real-time services.
With simple governance rules and ongoing oversight, a “non-neutral” network can unlock new, real-time services without harming general best-effort traffic and preventing any potential anticompetitive consequences. There are many possible ways to contain the possibility of abuse of traffic differentiation, ranging from relatively light oversight against unfair or deceptive trade practices from the Federal Trade Commission (the regime in place today) to a strict ban on paid prioritization. The right answer lies somewhere in the middle, with an expert, independent agency like the FCC overseeing prioritization arrangements on a case-by-case basis, informed and constrained by antitrust principles.
Traffic differentiation should not be only available to a select few firms. These services should be offered openly, ideally through a transparent API. No one should feel forced to purchase prioritization—or be prevented from doing so. Regulators should go further than simply banning exclusive dealing and require broadband providers treat like customers the same way. Such a rule ensures traffic differentiation is open to anyone on roughly the same terms, and is not limited to a select few firms.
With these simple rules, we can enable more intelligent broadband networks that better support real-time services while still preserving the fount of innovation that is the open Internet.