The interconnection battle has been framed as one between Comcast, Verizon, AT&T, and others on one side and Netflix on the other. A better way to think of it might be to put the eyeball networks on one side and the future on the other. What about the next Netflix (or any other business that Comcast, Verizon, Time Warner Cable, AT&T or CenturyLink view as competitive with their own)? What about all the other businesses that will be affected when Comcast finds the next network connection it wants to squeeze?
The problem is novel and complicated. But the M-Lab data is unassailable: this problem is harming consumers. Internet connectivity has been damaged for millions of users, for months at a time, with no consequences for the actors that caused the problem. If the FCC ever needed clear evidence of consumer and business harm, here it is.
In this context, ordinary customers can’t figure out what’s going on — remember that both Devan Dewey and Andrew Boegly are professionals, and they had enormous trouble — because these connections are far removed from anything a customer can touch or fiddle with. It is, in fact, odd that Netflix made deals directly with Comcast, Verizon, and AT&T during 2014: Netflix was just the customer of a network that wanted to interconnect. It should not be that someone doing the equivalent of “picking up the phone” to make a call — making material available online to be accessed by someone else — is unsure of what the quality of that call will be unless payments are made by entities not under his or her control.
Indeed, the US has long been against online “sending party pays” regimes internationally, because they raise transaction costs and create uncertainty for every element of the communications chain. In 2012, the European Telecommunications Network Operators’ Association (ETNO) proposed using a “sending party pays” method for charging for Internet interconnection. The United States government criticized the plan, with Lawrence Strickling, the Assistant Secretary of Commerce for Communications and Information, calling it a “solution in search of a problem.” Strickling noted that ETNO’s proposal would require complex negotiations among many players in order for packets to travel around the world. In contrast, traditional interconnection “works without requiring all of these parties to have a commercial relationship with each other or even to know everyone else involved in a given communication.” But in our own country, Comcast, Time Warner Cable, and Verizon, which collectively account for almost half of all fixed high-speed Internet access connections here (and, if you include Verizon Wireless’s customers, more than 165 million American customers) want those negotiations to happen and to be on their terms. The old system of engineer-driven interconnection is breaking down.
What’s going to fix this? It isn’t, strictly speaking, a net neutrality problem. Net neutrality is about how traffic is treated once it is inside an eyeball provider network. An operator could agree, as Comcast has, to net neutrality principles, but still control what gets into its network in the first place at interconnection points — which gives it just as much power to discriminate, albeit at an upstream point in its network.
No federal, state, or local government exercises any oversight over this handful of interconnection points. No Better Business Bureau watches over how your requests for data are being treated. The eyeball networks have unlimited numbers of levers to pull: they can discriminate by keeping interconnection doors narrow and by choosing with which networks to interconnect at all. And because each of the handful of large eyeball networks has such a large stable of subscribers, and because the number of physical interconnection points that matter is so small, transit networks have no choice but to comply with their demands. Users, meanwhile, have nowhere to go: Verizon, Time Warner Cable, Comcast, AT&T, and CenturyLink appear to be acting in concert.
The FCC has said that it is interested in interconnection and is gathering information on deals being made at the edges of the eyeball networks’ castle moats. Both sides argue the expense of expanding interconnection doorways: transit providers say that the costs are low — installing at one of the interconnection points a segment of fiber and an additional machine, costs the transit network would be happy to share — and eyeball networks say they’re high. But it turns out that eyeball networks are talking about the costs of running their entire network from the interconnection point to a household. They would like the transit networks or content providers to shoulder that full cost. (Which way the data is flowing makes no difference to the cost; according to the OECD, to carry 500Mbps in one direction requires a 655Mbps or 1 Gbps doorway, whether the traffic coming in the other direction is 10, 100, or 499 Mbps.) The transit networks, for their part, point out that the services and streams they are conveying to the interconnection points for admission into the eyeball networks are responsible for stimulating demand for high-speed Internet access being sold by the eyeball networks at sky-high profit margins. And they would be happy to bring the traffic right up to the door of the castle for free.
The FCC will find that the money amounts involved in these deals are low at the moment. It’s the naked threat posed to the future that is the problem: these charges could increase at any moment the terminating monopolies feel the coast is clear.
Who is right? It doesn’t matter. The eyeball networks have enough market power — remember, the only way to reach their subscribers is through their front doors — to require payment. And the terms they can exact over time are unconstrained by either oversight or market competition. This is not an argument that is won by persuasion or animated, in the end, by principle. It is based on an exercise of raw power.
Here’s what the FCC needs to do: first, it needs to use the M-Lab data as a prompt for deep examination into how packets are actually being treated at interconnection junctures with the terminating monopoly eyeball networks. Without this close look, the Commission will have no insight into how discrimination is being implemented by the eyeball networks.
Then the Commission needs to create rules for interconnection deals between the terminating monopolies and everyone else. These rules could require, for example, that the eyeball networks not be allowed to charge more for interconnection than it actually costs to set up the equipment needed to interconnect. Unless such rules are in place, other businesses and American consumers will continue to be squeezed.
Verizon, AT&T and Comcast are happy with the status quo.
“[A]rrangements concerning the interconnection of Internet networks and the exchange of Internet traffic have always been the province of voluntary, commercially-negotiated arrangements. . . . Applying prescriptive regulation in this complex and dynamic part of the Internet ecosystem is unnecessary and would have serious adverse consequences.”
“[M]ost commercial peering arrangements have a component that takes into account balance of traffic. When one side of a commercial peering arrangement sends significantly more traffic than it receives, the allocation of infrastructure costs described above gets skewed. If the sending party refuses to take steps to bring balance back into the arrangement, congestion can result. Since the entire Internet operates on a best efforts basis, severe congestion can only be addressed by a new commercial arrangement involving steps to either reduce the traffic flows, route the traffic in a more efficient way, invest to augment capacity, or some combination of all three. This is how the Internet has always worked— and it works very well.”
Comcast has argued that its merger with TWC should be approved in part because the deal would allow it to serve more business customers — the customers Cogent and Level 3 are now competing for — within its network, meaning that traffic would never have to travel outside Comcast’s moat.
“… Business customers and potential business customers stand to benefit from the expanded geographic reach of the combined company. In particular, Comcast’s larger footprint will allow it to serve more businesses on its own network and thereby likely (1) offer lower prices through reduced costs and reduced double marginalization, and (2) provide higher quality, more consistent service.”
The merger, if approved, will only make these interconnection problems worse and may, indeed, help drive Level 3 and Cogent out of the transit business altogether.
Note to FCC Chairman Tom Wheeler: Do something.