What the FCC Gets Wrong About Mobile
Last time, I explained what the FCC’s reply brief gets wrong about DNS. The brief argues that DNS is a network management function that ISPs employ to achieve more efficient routing, which saves them money. The correct, technically sound view of DNS is that it’s an information service that stores and forwards information from domain name owners to prospective users. DNS can be provided by third parties such as Google and OpenDNS, wholly separate from the ISP. Since Google has no incentive to spend its own money on a service that does nothing more than save money for ISPs, the FCC’s view is unsound.
The brief is peculiar in that it doesn’t offer a coherent and factual explanation of how DNS works or why the FCC believes it to be the Internet’s routing function, a role that’s actually carried out by IP with help from BGP. Rather, the only authorities the brief cites to support its view of DNS are legal ones. So the brief is long on legal precedent – it even throws in Justice Scalia’s non-binding view that DNS “is scarcely more than routing information” from his dissenting opinion in the Brand X case – but short on technical authorities such as Internet RFCs, text books, or academic papers.
Legal Precedent Does Not Clarify Technical Questions
It may very well be the case that the FCC’s lawyers thought they should stick to legal precedent rather than technical authorities because, after all, it’s arguing a court case. But there’s no law that says courts can’t consider technical analysis – they do this all the time in patent cases where the parties routinely and importantly rely on expert witnesses. But the FCC simply asserts an opinion with no support, and it turns out that opinion is factually deficient.
The reliance on legal precedent also undercuts the FCC’s claim that it needed to reclassify the Internet as a telecommunication service because the Internet has changed so much since the 2002 Cable Modem Order. The brief cites dozens of legal authorities that predate the Cable Modem Order, not just Scalia’s dissenting opinion. But if the Internet has changed so radically since 2002, why are these authorities still relevant? Some of them relate to legal principles that probably are still important, but the relatively uninformed opinions of judges and lawyers about the resemblance of DNS to call forwarding on the POTS network would not seem to be highly pertinent in the “brave new Internet” context. And none of these descriptions of DNS acknowledges the existence of Secure DNS (DNSSEC).
The portions of the brief that deal with caching make similar errors to the DNS arguments, generally confusing information with transmission. Caches – such as the Netflix server farms sprinkled around the countries in which Netflix does business – have an incidental effect on reducing communication costs for ISPs. But Netflix doesn’t build these facilities to save ISPs money by decreasing the distance information has to travel on ISP networks. Rather, they build these caches to reduce their own communication costs and/or to improve the quality of their transmissions. Caches don’t make transmission systems faster, they simply reduce the reliance of third parties on the transmission systems – such as they are – that ISPs maintain. A bit per second is a bit per second regardless of how far it travels.
Precedent Conflicts with the Brave New Internet by Definition
The brief argues that ISPs have new capabilities for traffic management that didn’t exist in the days of the Cable Modem Order, such as deep packet inspection, but this is not correct either. We had network sniffers in the 1970s that allowed engineers to view every last bit in packets moving along wires; all that DPI equipment does is replace a process that operated at low speed on low speed networks with a higher speed variation adapted to higher speed networks. The idea of differentially managing packets according to application requirements is as old as IP itself, as the original RFC for Internet Protocol from 1981 clearly shows. But when you muzzle engineering descriptions of networks in order to give more weight to the scare tactics of advocates and the loose analogies of confused judges, this is what you get.
Mobile Analysis Builds a Bridge to Nowhere
The portions of the brief that deal with mobile networks duplicate the errors in the FCC’s arguments for wired broadband reclassification and introduce some additional mistakes. The brief jumps through a number of hoops to demonstrate that mobile broadband is fully interconnected service that enables to the public to communicate with the public instead of a private network service. The brief seems to count on the inability of the court to distinguish cellular voice service from mobile Internet service. Voice certainly is a public network service that’s all about connecting people to each other.
But accessing web sites across the Internet is a whole different thing. The Internet is a network of networks whose main function is to connect privately owned and operated networks with each other. Each network in the vast mesh of networks that comprises the Internet only connects to those networks with which it has a bilateral interconnection agreement. So while it may appear to the naive observer that the Internet is some sort of a public commons, it’s actually private networks all the way down. If we have to choose one of public or private, the Internet comes down on the private side.
Web Sites are not Phone Calls
probably seems counter-intuitive because we’re accustomed to pretending that the Internet is fully interconnected and very public. While it’s highly interconnected, most of its connections are made by a very small number of firms that operate transit networks, such as Level 3, Cogent, and Tata. Small networks make one agreement with Level 3, and Level 3 makes thousands of agreements that enable it to get the small network’s traffic wherever it needs to go. So it’s all about private network to private network to private network.
There are also some big differences between making a phone call and accessing a web site. The FCC realizes this, so to paper over some of these differences it redefines the term “public switched network” from its traditional meaning, the telephone network, to a new meaning that seems to encompass all known and future modes of electronically-based communication. The redefinition closes the gap with respect to transmission between addressable endpoints, but it really doesn’t reach the fact that accessing a web site is not the same thing as making a phone call. Phone calls are bilateral, symmetrical, and personal, while web surfing is mainly unilateral, highly asymmetrical, and totally impersonal. Nearly every person in the FCC’s jurisdiction has the ability to make and receive phone calls, but not that many of us have our own web sites. Surfing the web is connecting to a privately owned and operated network, usually a for-profit commercial network, while making a phone call is simply an interpersonal communication.
The FCC Grants Itself Authority to Regulate Smartphones
The brief also argues that petitioner US Telecom is nuts to worry that the FCC’s re-designation of mobile broadband as a common carrier gives the agency the authority to regulate smartphones and the Internet of things. This comes down to the FCC’s fiction that ISPs sell “access to the Internet” when their actual service is more like “participation in the Internet”. This is a very subtle distinction on the wired broadband world because the roles of users and web sites are so different.
A web site is clearly part of the Internet, but the network in my home is not. That is, it’s not unless I do some things to overcome the limitations of putting all my devices behind a home gateway that makes them all share a single Internet address. If I have IPv6 devices and an IPv6-compliant router, all my devices are indeed part of the Internet and not mere spectators.
The same is true of our smartphones, devices that are not limited to calling out but which can accept incoming calls as well. And just as they can do that, they can and do serve up information to other services around the Internet. A home security system that can only call out and can’t answer inward calls is much less useful than one that can do both. And each mobile device plays a role in managing the wireless edge of the Internet in cooperation with other users and the private provider with which we contract. Smartphones engage in power management and traffic engineering, so that makes them common carriers by the FCC’s definition unless they’re specifically exempted. If an exemption is needed, that’s a clue that the rules leave something to be desired.
Impairing Public Understanding of Tech
It may well be the case that there’s policy payoff in pretending that the mobile Internet is just like voice calls, but I suspect that’s a pretense that leads to bad ends because it does violence to process and reasoned decision making. Even if the desired policy turns out to be correct, the technical argument for it is deficient. It comes down to the FCC’s admission that reclassification serves its policy goals, which in common sense English means they did it because they wanted to.
The FCC clearly believes the Internet is in jeopardy with careful scrutiny by the loving grace of the FCC, the would-be “referee on the field” who calls ’em like he sees ’em. The Communications Act doesn’t seem to direct the FCC to manage the Internet, so the agency redefines some terms, cuts some factual corners, and finds the authority to get involved in interconnection, service definition, and billing issues.
While these move may very well pay off in the distant future – they could motivate Congress to update the Communications Act – their shaky technical foundation worries me. The Internet is such a big part of modern life that it behooves all of us to understand it better, and on that score the FCC’s creative deconstruction of the Internet is not helpful.