ブロードバンド時代の制度設計

The revolution of the Internet has come. I come from a part of the United States, Silicon Valley, where we fear it has gone, it has disappeared. This extraordinary promise of innovation and creativity, which captured the imagination of the United States and the world in the 1990s now seems to have past. The question is why?

Just as surprisingly as this revolution occurred it has disappeared like a shooting star across the sky. We had no understanding of its source and now we have no understanding of its demise. But in my view the source of this extraordinary change and the source of its demise are both easy to understand if we know where to look. And our problem in particular in the United States is that an ideology blinds us to understanding the source of this great opportunity. My hope, in the few minutes I have this morning to talk to you about this architecture of innovation, is to first describe to you what it was about the Internet that made this transformation and creativity possible and then to describe what changes are occurring right now that have already and will continue to undermine this infrastructure and the potential that it offered the world for something different. My aim is to show you something that made creativity possible and the changes that now take that away. But to do this and make this understandable I want to first introduce two relatively simple ideas that I will use to help describe what it was about the Internet that made it distinct, and what it was about the changes we have now seen and will continue to see that take this distinction away.

I said I want to describe two concepts, both will be familiar but I want to describe them in a way that will make it easy for you to see the argument that I develop about the nature of what the Internet was. The first of these two concepts is the notion of layers. Computer designers and architects describe the Internet as comprised of seven different layers. I am a lawyer, so I have to simplify that to something more understandable for the limited minds of the legal profession. I want to describe three layers that we can imagine constitute the communication system both of the Internet or of any communication system. The first of these layers we can call the physical layer and it here describes the computers and the wires that link the computers that we understand as the Internet. On top of this physical layer there is a layer I will call the code layer, which is the logic that determines how the computers and the wires that build the Internet interact. On top of the code layer is a content layer, by this I mean simply the programs and the music and the web pages and other content that gets served across this network which is linked according to the logic of the code layer, pulling together the wires and the computers of the physical layer.

Now these three layers interact and together they produce the environment or ecology that describes the Internet. That is the simple concept of layers. Here is the concept of resources; I want to distinguish between free resources and controlled resources. By controlled resources I mean both those controlled by the government in the classic Soviet model of the government deciding who gets access to what resource and under what terms. But I also mean the control effected through private property of individuals who get to decide how resources get used. The essence of a controlled resource is that some actor has the power subjectively to determine under what terms a resource is used. I want to contrast these controlled resources with resources I will call "free". Now free resources are those outside of any control, for example, the air of most of the world is free in the sense that anyone gets access to it without control. But it also includes resources held in what the Anglo-American tradition called "a commons." Resources that are made available to people generally and equally without discrimination not necessarily for a price of zero but whatever the price is for that resource it is neutral among uses of the resource. So in the words of Richard Stallman, founder of the Free Software Movement, "Free" here means not "Free Beer" but "Free Speech". Not a resource that is available at zero cost, but a resource that is available without the permission of someone else, outside of the control of a separate entity.

Here are some examples then of communication systems that have each of them a physical, code and content layer, and each of them allocate these resources at each of these layers either as free or controlled. At one extreme is the example of London's Hyde Park Speakers Corner, a place where people gather every Sunday, with English lunatics screaming at each other about matters of who knows what. This is what the English speak of as "Free Speech." It is a context of political or social discourse which is important to the English and an oddity perhaps to the rest of us. But this communication system has a physical resource which is Hyde Park, and in the sense I described it that resource is free, it is a commons. And the code or logic that determines how communication occurs there is the English language and that language too is free, it is a commons. And the content which is uttered by these lunatics in Hyde Park on Sunday mornings, this is stuff no one would want to own, it too is free. This is a communication system which is free all the way down.

Now contrast this with the famous Madison Square Garden in New York City. Madison Square Garden is a physical resource like this building which is controlled, it is owned by the Madison Square Garden Corporation. So access to that resource is determined by the judgment of the owners of that resource. But in the context of certain uses of Madison Square Garden, for example, a political convention, again the code layer determining how people connect there is the English language, the American version at least and as English is generally free. In a political convention the content, the stuff that is uttered here in this convention hall is also free. This is a communication system controlled at the bottom and free in the middle and top.

Compare this to AT&T, the American Telephone and Telegraph Company, prior to its break-up in 1984, the monopoly-controlled telephone company in the United States. This was an architecture, the telephone wires and the physical telephone connected to this network, which was controlled, AT&T owned that equipment. The logic determining what could run on that network was also controlled. There is a famous story of Paul Baran in 1964 who had invented the concept of what we would call "packet switching", the core concept behind the Internet. He proposed this concept to the Defense Department and they in 1964 proposed this network design to AT&T. AT&T told the Defense Department that this network design, very much like the Internet, could never possibly work, but more important even if it could work, AT&T said "We are damned if we allow the creation of a competitor to ourselves." AT&T's power in this pre-break-up monopoly, was a power over the kinds of innovation that could run on their network and that was because the code layer of that network was controlled. But the content that was uttered on this network, the people speaking from both sides using the telephone, that too was free. So the telephone network is controlled at the physical layer, controlled at the logical layer, free at the content layer.

And now cable television in the United States and in most of the world; this too is a physical architecture that is controlled. The wires and equipment that makes this system run is under the control of the owners of the cable TV system, the logic that determines what is run on this network is also controlled by the systems that determine what gets selected for what channels. And finally, the content that decides what gets run on this network is controlled in the sense that only with the permission of the content owner can one broadcast certain content across this cable television network.

I contrast these four examples of communication systems, not to say one is necessarily better than another, if I had to spend my life either watching cable television or in Speaker's Corner London, Hyde Park, I am not sure which I would choose; but it is at least a tossup between the two. My point is not that ones to the right are better than ones to the left or vice versa, but I want you to see this range of possibilities so as to begin to elicit intuitions about the opportunity for innovation that each of these architectures permits. Because what distinguishes these architectures is the extent to which innovation about the use of this communication system is de-centralized from owners of the communication system itself.

Let us think about the content, the architecture of the original Internet. People speak of the Internet as if it is a "Free" communication system but according to this layout of the architecture it is not in the Speaker's Corner sense absolutely free. The physical layer of the Internet, the computers and the wires that connect them, is controlled for the most part as private property but some of it is government property. Regardless it is property that is voluntarily committed to this network and could be removed at any time. At the content layer, the Internet had content that was both controlled and free, controlled through copyright law, but an extraordinarily important part of the content layer is the range of free and open source software that was made available to users and developers on the Internet without the control of any particular user; devoted in this sense to free resources.

But the critical feature of the original Internet, the feature that distinguishes the Internet from every communication system before it, is at the code layer. Where the code layer of the Internet, the protocols that defined the original code layer, are in the sense that I am describing "Free". They describe a resource that is open for anyone to get access to and use according to their choice and this design was a chosen feature of the original Internet. And the description of the particular design they chose as articulated by network architects to Jerome Saltzer, David Clarke and David Reed was an end-to-end design. Now an end-to-end design says "build any communication system such that the intelligence in the system is vested not in the network itself but in the ends or at the edge of the network in the applications that drive use of the network." Design the network, as David Isenberg says, "to be stupid and make applications and uses at the edge of the network, smart." Simple network, smart applications, describes the principle we can call "end-to-end".

When network architects first described this principle and applied it to the context of the Internet they did so because though they were the smartest network architects the United States had to offer they were extraordinarily humble about the uses of this network. They knew they had no idea how this network would be used, so they designed the network so that it could develop and evolve without the central control of the network owner. They designed it so the network would be as simple as possible so that innovation and growth could occur at the edges and regardless of what was developed at the edges it would all run on the network. It was a commitment to an opportunity to innovate and create for this network that could not be controlled by the network owner.

They chose this design because they had no idea what the network would be used for but their design has an extraordinarily important competitive corollary. The competitive corollary is that the network owners of a network running an end-to-end designed network have no power to decide which applications and content will run on the network. What the network runs, which applications it serves, what kind of content gets fed across it is determined not by the network owner but by the users of the network at the edge of the design. Both who wants what and who offers what is a decision made not by networks but by users, and the consequence of that is that innovators who decide they want to develop for this network know that if their design is desired by users of the network then the network will run it. There is no intermediary who decides whether this use is consistent with the business model of the network owner.

So back to the story of Paul Baran in 1964, when AT&T could say to the Defense Department "We are damned if we will allow the creation of a competitor to ourselves"; AT&T could say that because they had a network that enabled them to choose which innovations would be allowed. In that network, innovators like Paul Baran knew that their innovations for a telecommunication system would be adopted only if they were consistent with the business model of AT&T, if not, then they had no opportunity to be adopted by the dominant provider of telecommunications. But the Internet embraced a fundamentally different constitutional value, and that value is that the network has no power to decide which uses will be allowed. So the Paul Baran's of this network know that there is no AT&T that gets to decide what innovations will be allowed under the architecture of the Internet. The innovations that will be allowed are just those that the market supports and the market here means the most diverse and diffused collection of users of this network that we can imagine. This is an architecture for decentralizing the decisions about which innovations will be allowed. And therefore an architecture for destroying the opportunity for large market players to behave strategically with respect to innovations that they do not want.

"End-to-end" means no one controls the right to innovation, which is therefore free. It exists in what we could describe formerly as an innovations commons; the right to innovate exists in the commons because the architecture of the network commits that right to everyone equally. That is the meaning of end-to-end. For five years we saw this architecture inspire an extraordinarily broad range of innovation that no one conceived of, no one imagined, supported and developed not by the largest network providers, not by AT&T, or America Online even. But developed by people who had no necessary connection to these original network control decisions. The idea of Internet chat gets born through a technology, called ICQ, that makes it popular, developed not by some insider at AT&T but by an Israeli who develops this outside of any existing network and is able within a couple of years to sell this technology to AOL for US$400,000,000 (I hope in cash not in stock).

The decision to develop the World Wide Web, was not a technology developed within computer network companies but a technology developed by a researcher at CERN research lab in Switzerland. A researcher could develop the protocols for the World Wide Web and implement them on top of the protocols of the Internet, because the protocols of the Internet were committed to an end-to-end design. He could therefore layer the protocols of the World Wide Web on top of the Internet and very quickly that innovation would take off. As network architects Saltzer, Clarke, and Reed wrote in 1993, "had the Internet compromised the commitment to end-to-end as many argued it should have in the 1980s, then there would have been no possibility for the World Wide Web to be built on top of the Internet".

We have seen these innovations that were enabled by an architecture that committed innovation to a commons. Now we are seeing changes to that environment, and it is my claim that these changes will fundamentally destroy the opportunity for innovation and creativity that the original Internet produced. In some sense the changes I want to describe here should have been completely predictable to anyone who believed that rational market actors will behave rationally to technological changes which threaten them. Because the set of changes I want to describe now are changes that are being brought about to permit those who had the most to lose from the emerging architecture of the Internet, to protect themselves from the innovations that this emerging architecture of the Internet would present.

This is a story of how old interests use their power in government and the market to protect themselves against innovations presented by the new. These changes I want to describe are occurring at both the physical and the content layer. There are changes at the physical layer which are corrupting the code layer and there are changes at the content layer that are likewise corrupting the feature of the code layer. The net result of these changes will be to transform the nature of the Internet in a way which will undermine the promise of innovation that this network produced.

First, changes at the physical layer. It was a completely accidental regulatory history that led to the Internet's birth. The Internet was born on a physical layer that was the telephone system. It was born just at the time when the American telephone monopoly lost favor with American regulators. It was born just at the time those regulators broke up the telephone monopoly and told the monopolists that they were not permitted to discriminate in access to their networks. So after the break-up of AT&T in 1984, competitors to this telephone monopoly increasingly thought that the development of this alternative network of communication, the Internet, would be promising. They had a right supported through law to go to the telephone companies and demand that they grant access to their networks to serve Internet service providers that made it possible for ordinary people to connect to this network through the telephone system. Regulation guaranteed competitors access to this network, and that guarantee of access was what inspired the explosion of Internet service providers (ISPs) around the US that increased competition to lower the cost to getting access to this network such that the network could take off.