Does the Internet need “governance”?

It’s remarkable to me that there are now two powerful agencies fighting to “govern” the Internet – the ITU and the FCC. On any given day, it’s hard to tell whether they are on the same side or different sides. The ITU process apparently began in earnest with the World Summit for the Information Society (WSIS) meetings, where the concept of “Internet Governance” became an urgent goal. The FCC process began when incumbent Internet Access Providers (IAPs) argued that “Net Neutrality” was a stalking horse for government control and definition of the Internet, followed by calls for regulatory definition of the Internet as a “Broadband Service” through the Regulation of “Broadband”. I recently signed a letter referring to the inconsistencies between the two efforts, which threaten, when combined, to destroy the whole idea of a “network of networks”, replacing it with a “vertically integrated service” concept, in the quest to “govern” something. But both efforts seem to have the Internet wrong in different ways.

As I’ve noted, the Internet is not Broadband, in my post What the Internet Is, and Should Continue to Be. But the FCC wants to view it as a “service” because for historical reasons, the FCC bureaucracy is organized around the idea that every activity or application of any sort in communications is a “service” that stands on its own. The Internet is a unification of all communications capabilities, so it just does not fit into the vertical integration idea that the FCC promotes by the structure of its regulations. (a reading of the enabling legislation does not require organization around “discrete services”, by the way)

The ITU also seems to be focused on “defining the Internet” as a “service”, but they focus on trans-national issues as well. In the US, the FCC does not deal with international communications – that is the province of the US State Department, which deals only with governments and quasi-governmental organizations, not with citizens of any countries. Though certain powerful multinational communications carriers are granted a seat at the table, largely because a number of countries including the US, have privatized their communications transport industries.

What is “defining the Internet” about? Well, largely it is about creating something to “govern” by inventing a governable entity, based on a lot of discussions with “stakeholders” [note: I am not considered a stakeholder because I represent myself. The technical language that defines stakeholder in both the FCC and ITU is someone who “represents an interest“, where interest is a governmental agancy, a corporation, or an organized interest group dedicated to influencing legislation in the interests of its members.The Internet users themselves are not interests]. And they are trying to define it as a service that is provided by a “provider” who owns or otherwise controls the medium. In other words, the assumption is that the Internet is a “vertically integrated” concept, that starts with applications, and is supported by a variety of gear that the “service providers” pay for, and resell to users in the form of services. This gives them a thing to “govern”.

This is attractive to bureaucrats who seek power and control over communications activities, whether the bureaucrats are in governments, international quasi-governmental agencies, or corporations. The move is to define the activity, and then limit the activity to a particular physical resource (wires, fiber, switches, gateways, spectrum property rights, …), and then control from the bottom. This paradigm of “governance” by creating property rights in physical media and then controlling all services built on that property is extremely attractive, and has reached full flower in the POTS and radio communications arenas.

But as I began, the question is: does the Internet need governance? By design and history, the answer is no.

The Internet was designed as a “network of networks” that could easily extend across all networks, merely by finding a way to transport Internet Protocol Datagrams (IP datagrams, or IP packets) across each network, whereupon a gateway (switch or router that understands IP datagram addressing) then can forward the IP datagram towards the eventual destination. Since all destinations and sources have IP addresses, the Internet Protocol and the gateways provide sufficient glue to create a universally connected network of networks.

This design avoids the need for any governance whatsoever in the delivery of packets. Further, the design is such that the content and intent of the datagrams need not play any role whatever in the gateways’ function. Only the IP Datagram “header” is used to make decisions about where the datagrams go. Part of routing the datagrams is the ability of the gateways to decide what route to take to deliver the datagram to the intended destination. But again, no global “governor” is needed to carry out the function efficiently – as the network of networks grows, a distributed algorithm for routing is both more resilient and more effective at getting packets to where they need to be.

Since content plays no role in Internet delivery, encryption of each datagram’s content may be used to further protect and to authenticate content against forgery. A key part of the Internet’s design was and is the ability to carry encrypted content for this reason – it prevents malicious tampering and reading of datagrams, up to the strength of the encryption algorithm and the key management maintained by the source and destination.

It is this ability of the Internet to be a universal network of networks that does not depend on applications that has led to its ability to serve as the “lingua franca” that spans international and corporate boundaries, facilitating any application that wants to use it, and also incorporating any underlying technology for communications – starting with dedicated digital circuits and voice-grade switched lines using acoustic encoding (so-called “dialup”), and now including fiber, cable TV coax, wideband radio, mesh networks, etc.

So the Internet is not an application or “service” in the sense that the ITU and FCC would like to define it. It is not “Facebook + Google + Instagram + The Cloud + email + Twitter + Amazon + iTunes + Alibaba” – an amalgam of current popular services that happen to exploit the universal openness of the Internet.

Nor is it Broadband or LTE or GSM.

This is why the “network neutrality” discussion, framed as “who will govern the Internet” is wrongheaded from the beginning.

As Ive noted, the Internet “needs a little help from the Law”. But the key point here is that law is not the same thing as governance. Laws are rules that humans (and “persons” like corporations) must obey, or be punished. Not all laws come from governments. There is a whole body of “common law” that is generally accepted, transcending government. One such law is that you cannot steal a package that you’ve agreed to transport from point (a) to point (b). That is true whether or not there is a “contract”. It’s just not done, and courts in any jurisdiction, no matter what the government, will hold to that principle.

So reading and benefiting from a private communications that you happen to be carrying as part of the Internet should be covered by this standard principle. We don’t need the Internet to be “owned” as a whole, or “governed” as a whole to prevent that or to discipline those who might do so.

Similarly, discriminating at a hotel based on the color of some guest’s skin is equally noxious. There are those who think all laws should be based on absolute property rights who struggle to find such ideas acceptable – usually by defining people as non-persons due to their forbears’ genetic makeup. But in a modern society, we know that there is no basis for such discrimination.

There is a tendency to blame the Internet for the kinds of communications that go over it – and to try to hold the Internet liable. But the criminal behavior that happens over the Internet is not caused by the Internet transport of packets. Again the idea that the Internet is somehow a service is based on a fundamental confusion. Should we blame the English language (or the Pashto language) because people can conspire to commit crimes by speaking in English (Pashto)? Should we blame a culture’s literature and newspapers for the behavior of individuals who belong to that culture?

Trying to conceive of the Internet in terms of “governance” reflects a peculiar redefinition of what the Internet is about.

The Internet is a form of universal glue. It’s built by those who use it, and based on a design concept that allows a network of networks to scale to any size on any technology that can carry IP Datagrams.

What the Internet needs, however, is some help from the law. The help is required largely because governments create or subsidize monopolies. Examples include radio spectrum rights (you cannot get the right to operate a transmitter in the US or a receiver in countries like the UK without a very restrictive license that bars most modern communications techniques other than those of a small set of “incumbent” providers), and local “franchise rights” created and maintained by local and national governments (RCN was not allowed to build out Fiber in Philadelphia, the corporate home of Comcast, by a mayoral decision based on the claim that it would “cost jobs”).

The Internet can run fine across these monopoly platforms, but the temptation of some of the monopolies is to claim the right to muck with Internet packets – and this is not a theoretical claim. It is at the core of behavior that has been documented, including products from Phorm, NebuAd, Sandvine, Ellacoya and others that are designed to read all IP datagrams to analyze content, modify content, act as a “man-in-the-middle” to control connections unbeknownst to the endpoints, etc. Those companies are doing great business selling to access providers the tools to exploit what is inside of IP Datagrams, in most cases without disclosure, and if disclosed there is only a mention of the possibility in a Terms of Service, and maybe an obscure “opt-out” mechanism that can be offered when the exploitation is discovered.

If the state grants such monopolies, the state must be responsible to police those monopolies’ actions. And that’s one place where the Internet needs a little help from the law.

One could argue that the Constitutional Protections in the US for Free Speech and Free Assembly only protect against “government action” to block free speech – that companies who interfere with speech and assembly on the Internet do so privately, and therefore outside the purview of the US Constitution. But that is clearly wrong for a simple reason: the government created the monopolies! So the government is responsible for the curtailment of free speech and free assembly by its monopolies. That includes monopolies at the level of Towns, Cities, States, and other jurisdictions in the US – the constitutional rights bar those governments from mucking with speech and assembly rights. And this principle goes beyond America – many (if not most) countries guarantee freedom of speech and assembly, and most, if not all, countries grant monopoly rights to communications carriers.

This could be easily solved with a simple law: any company that handles Internet datagrams may not read or modify the content, nor infer intent or meaning for the purpose of deciding what datagrams to deliver or to not deliver.

That’s a pretty simple principle, and it happens also to be the design principle behind the Internet, and what has made it work.

If a Cable TV company chooses not to offer Internet service at all, that’s fine. Let them. They then would not have a “franchise” right for Internet service, and someone else who chooses to offer full Internet service could enter the market, which is awesomely large! There’s no risk there.

Similarly, if a wireless company chooses not to offer Internet service at all, great! Again, they would be wasting the value of their monopoly spectrum, and someone would find a way to offer Internet service.

But the law would merely exist to protect the simple rules about IP datagrams – no peeking, no changing, no routing some but not others based on content.

We might not even need the law if the governments would get out of the business of granting monopolies, as I have argued is possible (and needed) in the case of digital networked radio technologies. The argument that spectrum rights are needed for radio communications to function is technically wrong in a fundamental way. In fact, we would have vast improvements in wireless capability if we were to take advantage of the ability of digital techniques (modulation, sensing, cooperation, and interoperation as a network of networks) for radio. The major block here is that the current incumbents control the regulatory framework, because they like monopolies given out by the government. However, separating the necessity of monopolies from the question – the law can easily say that the monopolists that offer Internet access and transport must not peek, modify or discriminate, as a condition of holding the monopoly.

We might not need the law if we could adopt a universal framework for encryption and secure routing among all the glue parts of the Internet, such that there would be no ability to peek, modify, or discriminate. I find this less likely to happen, because it would require a significant effort on the part of vendors and applications that use the Internet to ensure that all these parts get built and implemented widely.