Application-Centered Internet Analysis

Timothy Wu
1999 Virginia law review  
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact Virginia
more » ... w Review is collaborating with JSTOR to digitize, preserve and extend access to Virginia Law Review. HERE is a now-standard debate about law and the Internet. One side asserts that the Internet is so new and different that it calls for new legal approaches, even its own sovereign law. The other side argues that, although it is a new technology, the Internet nonetheless presents familiar legal problems. It is a battle of analogies: One side refers to Cyberspace as a place, while the other essentially equates the Internet and the telephone. In my view, these two positions are both wrong and right: wrong in their characterization of the Internet as a whole, yet potentially right about particular ways of using the Internet. The real problem is that both sides (and indeed, most legal writing) rely on a singular model of the Internet. They take one way of using the Internet as a proxy for the whole thing and conclude "the Internet this" or "the Internet that." In earlier times, this simplification worked. Then, the Internet was new and less diverse. But today, most noticeably for purposes of legal analysis, the singular model is failing. In actual usage, on which legal questions usually turn, the Internet does not generalize well. The Internet is only the genus, the application, the species; applications can and do vary dramatically. To the user, the Internet comes in many incarnations-email, the World Wide Web, ICQ,1 and more. A singular model of Internet usage has become too small to capture the dramatic diversity of today's Internet. * Clerk to Justice Stephen Breyer, Supreme for extremely helpful comments and suggestions; and the members of the Dirksen Paper Talk Group, where this Essay was first presented. I For a description of ICQ, see infra note 103. 1163 1164 Virginia Law Review [Vol. 85:1163 Accordingly, I would like to suggest an upgrade. For most purposes, I think we ought to discard the old talk of the Internet as a whole, for the whole Internet is rarely an appropriate level on which to generalize. Instead, legal thinking can better focus on where the variation that is apparent to the user is actually found: the application layer above the Internet's basic protocols. We need, I think, to focus on the user, not on the network, and that means legal analysis that begins with the application. What's the difference? This seemingly technical point matters because the Internet by its design allows-even encourages-great diversity above a few basic standards. The "end-to-end" design of the Internet delegates the power to code function to the point nearest to the user: the application. As a result, nearly everything that "counts" about the Internet from a legal standpoint is a function of the particular application at issue and not of the basic Internet protocols. Since applications actually drive Internet usage, they ought also drive legal analysis of the Internet, yielding nuanced rather than stereotyped results. The crucial point to understand is that the Internet was expressly designed to put the application in charge. (Importantly, by this I mean the application, broadly conceived-the programs themselves (e.g., email, telnet, browsers, etc.) and their associated protocols; in other words, everything above the basic Internet standards.) The Internet, like many networks, has a layered architecture. That is to say, all the tasks necessary to communicating via network are divided among several functional layers, and the programs residing on these layers cooperate in standardized ways. Applications and their associated protocols occupy a layer above the basic Internet protocols that supervise basic data transmission. And so we can see that the designers of the Internet had a real choice about where to place the functionality of the network-how much freedom to give to the application and how much control to maintain through the low level protocols. The monumental choice-expressed famously as a design principle in Jerome Saltzer, David Reed, and David Clark's "end-to-end" design argu-ment2 was to make the basic Internet protocols simple, general, 2 Jerome H. Saltzer et al., End-to-End Arguments in System Design, 2 ACM Transactions in Computer Systems 277 (1984), reprinted in Innovations in Internet- 1999] Application-Centered Internet Analysis 1165 and open, leaving the power and functionality in the hands of the application. As a recent comment by the same authors explains, "Moving functions and services upward in a layered design, closer to the application(s) that use them, increases the flexibility and autonomy of the application designer to apply those functions and services to the specific needs of the application."3 The impact of the end-to-end design principle, embedded in the architecture of the Internet, is crucial to any analysis of the Internet, and legal analysis is no exception. To understand this point, just think of the "network" of appliances in your home: They all use the same standard of electricity (the basic protocol), but then widely differ in what they offer the user. A television offers something quite different than the power saw, even though both use 110 volts of electricity. This is the result of a deliberate choice. The design of the electricity "network" puts most of the power to decide functionality in the hands of the appliance designer. The Internet, conceptually, is not all that different. Contrast this with the telephone network, where nearly everything that matters about the telephone comes from the basic standards to which all telephones adhere. The difference between these networks is the result of a deliberate and important decision, and one that cannot but have a decisive impact on the legal analysis of any network. The Interlude between Part I and Part II explains these points in greater detail. working 195 (Craig Partridge ed., 1988) . "End-to-end" arguments are referred to as such because they recognize that a class of functions can only be completely and correctly implemented by the applications at each end of a network communication; hence, delegation of this function to lower protocols will generally be redundant. Put another way, end-to-end design arguments recognize that building complex functions into lower levels of a network implicitly optimizes the network for one set of uses but may then increase the costs of the network for uses that were unpredictable or unknown at the time the network is designed. The end-to-end design principle is featured prominently in the helpful summary of "Architectural Principles of the Internet" in RFC ("Request for Comments") 1958 (visited July 13, 1999) ; see also infra notes 74-77 and accompanying text (discussing the end-to-end design principle). I David P. Reed et al., Active Networking and End-To-End Arguments, 12 IEEE Network 69, 70 (1998). Virginia Law Review [Vol. 85:1163 The purpose of this Essay is to encourage a legal analysis that is more cognizant of the effects of the Internet's network architecture. That architecture provides, and even encourages, a rich and diverse universe of possible applications, foreclosing simple generalization of the Internet as a whole. For this reason, much legal analysis ought usually begin at the level of the Internet's individual applications, and not at the level of "Cyberspace." What this ultimately means is an analysis that focuses on the user, and how the Internet actually appears to the user, rather than an abstract focus on the network as a whole. This richer focus on Internet usage makes things a little more complicated. Therefore, the second purpose of this Essay is to suggest coherent ways to think about and to classify the parts of the Internet-to dissect in legally relevant ways the universe of existing and possible Internet applications. Sometimes it makes sense to look at applications individually; other times, applications can be grouped by functional characteristics or by adherence to certain protocols; and in certain cases every application that adheres to the Internet's standards will be affected similarly, making an analysis of the whole Internet reasonable. I suggest two tools for grouping applications in this Essay that correspond to two main areas of Internet law, though the field for grouping applications is open. Two main areas where the law and the Internet have met-proxies for public and private regulation questions, respectivelydemonstrate the great difference that an application-centered analysis makes. Part I considers the First Amendment analysis of the Internet, a discussion that stands as representative for any issue of public regulation of the Internet. Part II brings applicationspecific methods to the now-historical debate over whether some form of strong self-governance makes sense for the Internet. This discussion, in turn, is representative of questions of private, contractual regulation of the Internet. 1999] Application-Centered Internet Analysis 1167
doi:10.2307/1073968 fatcat:qnuzppr2k5fxbbnhf5zcpoj65i