TMCnews Featured Article

It’s a simple question and it can be read two ways. One way is to ask just what constitutes the Internet, or makes something a part of the Internet. The other way is to ask whether and why it matters. Those are important policy questions in today’s world, where “Internet” is still a magic word, beloved of many policy makers but still widely feared.

The first question is critical. Defining just what the Internet is not simple. Justice Potter Stewart once explained that he couldn’t easily define obscenity, “but I know it when I see it”. We think we know what the Internet is, but to define something is to draw a limit as to what the word does and doesn’t refer to. Examples don’t count; they are just illustrations. We know that the Internet today contains millions of computers in a worldwide network-of-networks that uses the TCP/IP protocol stack to provide a range of applications, but the Internet isn’t defined by its protocols or applications any more than “horse” is defined by the Kentucky Derby or Alpo.

In fact, the Internet we are using today is not the final product. It’s a prototype, a lab experiment run amok. It’s a complicated experiment, one that goes beyond computers and cables and fibers and into whole methods of organizational development, standardization and financing. But it’s not the final product. That only emerges after answering the question with two meanings, what does it mean to be Internet?  A real definition would be useful, and exactly how that definition is crafted could have many implications. So here’s a definition that is no more or less encompassing than necessary:

This definition is intentionally terse, a distillation of the concept to its essence. It simply defines a business model for interconnecting networks. It does not specify a protocol. TCP/IP has been in general use since the early 1980s, but it’s running out of steam, and a replacement protocol or two may come along soon. Calling something “IP” shouldn’t convey an exclusive right to be considered as Internet. By the same token, use of an Internet Protocol such as IP does not make something Internet. IP is already used within the telephone network and other non-Internet applications. Certainly the telephone industry’s IMS (IP Multimedia Subsystem (News - Alert)) is anything but Internet! IP is a tool. It’s a tool designed for the Internet, but not all usage of IP is Internet.

The definition does not include any reference to the current global big-I “Internet”. That is an Internet, but it need not be the only one. Today’s prototype demonstrates the Internet business model, but even now, some of its participants are not really happy about it.

The definition does not include a reference to the Internet Engineering Task Force, the Internet Society, ICANN, the United States government, or any other organization. They are all participants, and they all helped create it, but their continued participation doesn’t define the model. It’s not an exclusive club. Anyone can play.

What the definition does specify: It is voluntary.  This is a very important part of the business model, one that is unexpected in the telecom industry, which is characterized by historical monopolies, but common in other forms of commerce.  Indeed, Internet is basically what you get when ordinary commerce takes place in the form of transmitted information, rather than hard goods. 

Network operators don’t have to join in, or play by anyone’s regulations. An Internet is basically a set of contractual, usually bilateral, agreements. So if a network operator doesn’t want to allow another one to connect, it doesn’t have to. It doesn’t have to accept any traffic it doesn’t want, whether it deems it spam, malware, or simply excessive. It doesn’t have to pay anyone their asking price; it can just walk away – there is no mandatory interconnection between ISPs. There is no tariff with fixed prices open to all comers. And thus there is no bright-line distinction between network operator and customer, no formal definition of who is or isn’t an ISP. Peering – the interchange of traffic between ISPs at no cost, the privilege that distinguishes retail customers from ISPs – is not mandatory, but reflects the mutual benefit of the exchanage of traffic. And if that mutual benefit involves a little cash in either direction, fine. That’s what voluntary is about.

Contrast this to the traditional telecom industry, which operates the Public Switched Telephone Network (PSTN). That’s based on a regulated public utility business model, one that really goes beyond plain old telephone service (POTS). The PSTN has licensed service providers who are common carriers. They file tariffs which set uniform rates for all takers. Most PSTN providers are also all too happy to bill for every little transaction, an “annoy the customer” feature that festers in a monopoly environment.

PSTN Protocols are often specified by authorities, or by large Incumbent carriers who have the type of monopoly power that doesn’t occur in the Internet. There’s a bright line distinction between a PSTN customer (“subscriber”) circuit and an intercarrier connection. Interconnection between carriers is mandatory – they must accept each other’s traffic, but the rate they are allowed to charge each other is also set by regulatory authorities, via tariff or regulated contract. 

This isn’t really a bad business model, since the PSTN helps provide “universal service” in the sense of one worldwide network that almost anyone can belong to at a predictable price. It is well suited to dealing with monopolies and with activities where competitive forces are weak. But it’s not very flexible, and it is prone to both monopoly abuse and regulatory friction.

The distinction between Internet and PSTN thus does not hinge on whether the traffic in question is voice, data, video, or anything else. It doesn’t hinge on whether the switching is based on packets, circuits, or for that matter smoke signals. Those are just the tools. It hinges on the business model. The Internet business model, when it works, can be applied outside of the “big-I Internet”. And the PSTN can make use of the same tools, such as packet-switching protocols, without becoming Internet. The two are largely complementary. To some extent they compete, but they have different strengths and weaknesses, and both sides need to learn to work better with the other. The highly-competitive Internet model only thrives when given access to regulated PSTN carriage, which needs to treat it as bit-neutral customer payload.

So what else does it mean to be Internet, in that second sense, now that we’ve defined it? For one thing, it means that orthodox “network neutrality” is literally impossible to apply to ISPs, as that means involuntary interconnection. At that point the ISPs have been swallowed into the PSTN! 

The problem is that the Internet sometimes works so well, albeit by brute force, that people have started to use it as a substitute for the PSTN and even for cable television, yet another business model with its own regulatory baggage. Its voluntary, free-market business model allows it to be cheaper than the PSTN for some applications, since the PSTN lacks sufficient competition, and still heavily taxed and cross-subsidized in order to fund “universal service” and to achieve other political goals.  And no good deed goes unpunished. Even worse, some are calling for regulation of Internet content providers (web site operators), to regulate what they charge or to whom they must make their proprietary content available.

Not only are these ideas tantamount to regulating the press, but they’re like telling publishers what they can charge and to whom they must sell their wares. Internet is about content, not carriage. That’s why it is classified in the United States as an “information service”. A free press does not mean one in which all publications must be given away free-as-in-beer. Democracy is the right to own a press, not to dictate what others publish. And the lack of access from the subscriber to an unrestricted choice of ISPs – the current cable/telco (think of Pravda and Izvestia, the two national newspapers of the old Soviet Union) duopoly – is the real threat to network democracy.

The PSTN, in contrast, must be content-neutral; because that’s what its role, common carriage, is all about. We almost had a PSTN able to provide high-speed packet-switched data, video, and other services – that’s why ATM (Asynchronous Transfer Mode, a networking protocol still widely used within DSL access networks) was invented over two decades ago. But it failed because the billing-happy PSTN business model couldn’t accommodate it, and the Internet stepped in to fill the void.

The post-2005 US regulatory model, which treats the Internet as a vertically-integrated stack right down to the wire or radio waves that carry it, is literally self-contradictory. Something can’t be Internet and PSTN at the same time. It doesn’t work. The resulting regulatory friction is already proving to be incendiary. Yes, ISPs can be self-provisioned non-common-carriers, but the telecom industry can’t be treated as Internet just because it also provides retail ISP services. Indeed the public Internet only became possible because of regulations the FCC (News - Alert) had imposed well before 1984’s AT&T divestiture. One was a regulatory requirement that carriers allow sharing and resale of its lines. The other was a requirement that content-based (“enhanced”) services be treated as neutral payload, with PSTN carriers having to treat their own enhanced services the same as their competitors’.

Another thing that being Internet means is that there is no authority. PSTN operators are beholden to government regulators; the Internet isn’t. ICANN, the Internet Corporation for Names and Numbers, is merely a voluntary organization. It coordinates names and numbers on the “big-I Internet”, but ISPs are free to ignore them and pick IP addresses of their own choosing. Of course since interconnection is voluntary, other ISPs are equally free to refuse interconnection, or they can choose to create their own workarounds such as encapsulation or translation. This sort of thing is unlikely to happen much, but it provides a check on organizational behavior. ICANN is, in effect, elected by ISPs who choose to use it.

This lack of authority extends even more strongly to the IETF. This voluntary organization dates back to the days when the Internet was the private domain of the government and its selected partners. It uses a “consensus” technique to create Internet standards. Consensus generally boils down to how many people show up in the room. Organizations and countries don’t vote; even individuals don’t vote. The loudest position wins. This tends to favor large organizations who have a vested interest and enough geeks and money to make sure that they’re well-represented at meetings. It has not proven, however, to be effective in maintaining consistently high quality.

The IETF, and its overseer the Internet Architecture Board, lost sight of the ball about 15 years ago when they adopted IP Version 6 as the “next generation” Internet protocol. IPv6 fails almost every test for a proper next generation protocol. It does not really address multihoming, multicast, or mobility, three weaknesses of IP that have been understood for decades. It doesn’t handle inter-provider scaling and it lacks backwards compatibility. It is merely a costly, crude fix for a perceived shortage of IP addresses. But the IAB and IETF are unwilling to admit to having made a mistake, so their current answer is to shout louder, even as most ISPs and Internet users correctly ignore them.

So a new Internet may well evolve entirely outside of the purview of the IETF and IAB. It could take shape among service providers, equipment vendors and users. It could take place without even being noticed by the government or big operators. Or the voluntary standards bodies could catch up. Humans are social animals with an urge to communicate. And thus Internets will continue to be built.