As a follow-up to CZ’s post on Setting the Record Straight for The Economist Blog, I wanted to elaborate on why we believe it is appropriate to approach Internet policy as an ecosystem as opposed to focusing only on a particular layer in the TCP/IP model. Our view is premised on two concepts.
First, while the “layered” nature of the Internet from an engineering perspective is a useful concept, it has no meaning in the policy environment. The underlying premise of using the layers engineering concept is that the networks or physical layers of the Internet should be subject to regulation. But this notion is based on a rigid concept of the Internet. Often services offered by network providers incorporate many of the “layers” in order to offer the services. Even basic broadband Internet access incorporates higher layers of the Internet than just the physical layers. In point of fact, the layers are meaningless when it comes to defining regulatory policy.
More fundamentally, the concept of the Internet ecosystem that we and many others use to describer the way the Internet works today is based on the fact that harms to competition or consumers can occur at any layer of the Internet, and any policy that is narrowly designed to address problems only at the lowest layers will fail to achieve the objective of preserving an innovative, competitive and dynamic Internet ecosystem.
While the commercial Internet is still relatively young, the last 25 years have provided some instructive examples that demonstrate the need for an ecosystem approach to Internet policy. Professor Grunwald at the University of Colorado, Department of Computer Science released an excellent technical report in March that describes many of these examples and I highly recommend Dr. Grunwald’s report to anyone interested in technology policy.
One of the key points in Dr. Grundwald’s paper is that, “[T]he ability to limit access to Internet applications is not restricted to access networks.” And he continues to explain that, “Such restrictions can be enacted by many components used to access Internet content, such as the browser and services or applications within the Internet.” He then provides a few examples of components that have the potential to restrict access to content, limit choice, or enhance the delivery of selected content. Everything from operating systems to applications to browsers to search engines could certainly be engineered in ways that interfere with or even degrade consumer access to the Internet as Dr. Grunwald points out. It is even conceivable, Dr. Grunwald notes, that some players would have incentives to cause such interference.
So harms to users or competition can happen at any layer of the Internet. But my view is that such actions are unlikely to occur and if they did, they would be unsustainable. Why? First, because the Internet’s growth and advancement has been based on a collaborative, competitive, self-governing model. That sounds like an oxymoron but in order to deliver the services, traffic or applications that consumers want, packets have to go through many different networks, up through many layers of the Internet and into computers and back again. All players have an interest in making sure all of this works because everyone is either a consumer or a provider – or both – in today’s interconnected Internet.
Second, the Internet’s open protocols and interactive nature mean that little can be kept hidden for long. There are no “centralized” sources of knowledge or control on the Internet. Engineers, technologists, academic experts and company specialists all are involved in the Internet. Expertise is widely dispersed as is knowledge and access to networks and services. The Internet’s “self-governing” nature is a based on this widely dispersed knowledge base. If someone did do something that interfered with, degraded or undermined consumer access to services or use of applications, it would not be hidden for long. And public exposure will almost certainly force a change in any damaging activity.
None of this means that providing enhanced mechanisms for delivering traffic – whether network based or involving improvements in caching, storage or applications – are bad things. They clearly are not and many technologists and Internet gurus have made the point that the Internet was not designed to deal with all the services or traffic that we find available today, such as video.
But Dr. Grunwald’s paper makes clear that the “layers” approach has little relevancy to actual outcomes so policymakers shouldn’t try to make an engineering design principle the basis of a policy framework for the Internet. As CZ noted last week, we have suggested a way forward and will continue the dialog about an ecosystem approach to Internet policy.