Why Vox (and Others) Are Wrong About Internet Interconnection


By Steve Morris and Jim Partridge A recent piece on Vox argues that disputes over network interconnection are the “new battle for the future of the internet” and may prove more important than the current debate over net neutrality. While the tone, slick graphics, and level of detail in the story create the appearance of trying to provide a meaningful explanation of an important issue, the piece ultimately misses the mark in a number of significant ways. Figure 1 of the article attempts to show a highly simplified model of congestion. In this example, all traffic coming into the ISP network from the transit provider is treated the same and all the applications are degraded due to the congestion on that link. What’s omitted from this picture is that, in the real world, most ISPs have connections with multiple networks, including transit providers, content delivery networks (CDN), and some direct connections with content companies. The same is true of major content providers. In short, companies do not, or at least should not, leave themselves wholly dependent on a single transport provider.

Screen Shot 2014-11-18 at 3.08.23 PM

Figure 2 attempts to show a hypothetical paid prioritization arrangement where one content provider – MyFlix – pays the ISP – AT&Tcast – for prioritization so that its service is not hampered by congestion. Whether the diagram accurately portrays the prioritization described in the article is questionable, but the more important point is that it fails to mention that no American ISP offers such a service and most have expressly disclaimed any intention to do so. So Figure 2 is not merely hypothetical, it is pure fiction, at least for access networks.

Screen Shot 2014-11-18 at 3.08.36 PM

This error is compounded by Figure 3, which attempts to show that a direct connection between MyFlix and AT&Tcast, is “identical” to paid prioritization. This seems to be the article’s key point. But it is completely wrong.

Screen Shot 2014-11-18 at 3.08.46 PM

The attempt to equate direct connections, which some ISPs do offer, with paid prioritization, which no ISPs offer, fails to acknowledge the different effect that the two arrangements have on content providers. When MyFlix transfers its business relationship from the transit provider directly to the ISP, in Figure 3, not only does its own performance improve, but the other two content providers, YouBook and FaceTube, also benefit because the transit link is less congested. That is not the case with the paid prioritization example in Figure 2, where the prioritization benefits the company that pays but does not change the overall level of traffic flowing through the congested transit link. Based on the erroneous conclusions drawn from hypothetical examples, the article highlights the arrangement between Comcast and Netflix and concludes that regulation of ISP interconnection practices may be warranted. As with the hypothetical examples, this analysis is a vast oversimplification that ignores highly relevant facts. In particular, it fails to acknowledge that Netflix had dozens of options for delivering traffic to Comcast, but chose to limit its traffic to a handful of providers. Moreover, Netflix stuck with that decision even after it became clear that using these providers was affecting the performance experienced by its customers. The article also ignores the fact that moving Netflix traffic to a direct connection arrangement appeared to relieve congestion that other web users were experiencing.

"The attempt to equate direct connections, which some ISPs do offer, with paid prioritization, which no ISPs offer, fails to acknowledge the different effect that the two arrangements have on content providers."

In short, the portrayal of Netflix as a victim of Comcast’s hardball tactics isn’t supported by the facts and it ignores that the resulting direct connection arrangement is one that clearly benefits content providers and consumers by reducing the potential for congestion and thereby improving the performance of content and applications. Despite the flawed analysis, we do agree on one important point raised in the article – that trying to address these matters through regulation would be extremely difficult. As the author acknowledges, “[s]omeone needs to pay for the cost of these connections, and the fairest way to split the costs depends on many subtle factors, including geography, traffic patterns, and the relative size of the interconnecting networks. A poorly written interconnection rule could create a lot of work for lawyers without actually preventing abusive practices.” Indeed, attempting to overlay regulation on a system that currently transmits hundreds of millions of gigabytes (U.S.) of data daily through a web of thousands of contractual arrangements could lead to significant arbitrage and litigation. Internet interconnection practices are well-established business relationships that some want to drag into the regulatory abyss. Vox readers deserve a better analysis of the facts.

Steve Morris serves as Vice President and Associate General Counsel at NCTA. Jim Partridge serves as Vice President of Industry and Technical Analysis at NCTA. Image Credit: Vox Media