Law and the open Internet.

• Author: Candeub, Adam

1. INTRODUCTION II. ECONOMIC ANALYSIS DOES NOT PROVIDE POLICY GUIDANCE A. Is Vertical Foreclosure the Right Lens for Network Openness? B. Is Foreclosure a Reasonable Concern? 1. The Single Monopoly Rent Theorem, Vertical Integration Stories, and Broadband C. Does Characterizing Broadband Service Provision as a Two-Sided Market Further the Analysis? 1. Does Two-Sided Market Analysis Apply to Broadband Service Provision? D. Does Anyone Know Anything About Dynamic Efficiency? III. NETWORKS DISCRIMINATE, AND NETWORK ENGINEERING GIVES POLICY MEANS NOT ENDS A. Selecting Where and What to Build B. Configuring to Discriminate 1. "Best Effort" 2. "Quality of Service" C. Selective and "Fair" Interconnection IV. AN EMPIRICAL, NONTHEORETICAL LAW OF THE OPEN INTERNET A. BitTorrent-Comcast Controversy B. Cogent-Sprint Disconnection C. Madison River V. A JURISDICTIONAL POSTSCRIPT VI. CONCLUSION I. INTRODUCTION

   The FCC has recently issued a new set of Internet access regulations and policies: the Open Internet Proceeding Report and Order) This order culminates over a decade of intense policy debate on whether any rules, and, if so, which rules will work best to structure the Internet's component markets. Advocates of regulation frame it as an effort to preserve "net neutrality." (2) They fear dominant broadband providers, such as AT&T and Comcast, will use their market power in consumer markets unfairly, favoring Interact content in which they have financial interest. (3) They argue that the FCC should introduce strong regulation to defend the Internet--a place that teems with decentralized creativity--from the centralizing strictures of a glorified content delivery system. (4) Foes of Internet regulation fear that openness mandates might create burdensome government intrusion and reduce infrastructure investment. (5) They fear that the FCC's fitful fantasy of slaying anticompetitive dragons will impose new government controls that smother the real and innovative wonder of a free Internet. (6)

   The Open Internet Proceeding focuses on one issue: "last mile" anticompetitive foreclosure. (7) Broadband access providers, like AT&T, Comcast, or Verizon, might favor affiliated online content and disfavor competitors. For example, Verizon might block access to Skype, a competitor in telephony; Comcast might block or degrade online video downloads from Netflix, a competitor in cable television services; or any of them might accept money from Google to degrade Yahoo's traffic. The FCC Open Internet Order employs the normative principle of "reasonable discrimination" as the legal standard for Internet regulation, (8) following the FCC's analysis in the NPRM. (9) Applying this principle to how the network transmits packets of data, some phrase the rule as requiring that "like content" gets "like treatment" without discriminating based on the type of application or discriminating by the identity of the sender or receiver. The Open Internet Order applies this principle to broadband access providers. (11)

   This Article first shows that this regulatory approach and the attendant scholarly debate are fatally narrow. With their noses buried in the last mile, commentators and regulators do not account for the full range of tactics available to a network when it wants to discriminate. Internet networks--beginning with those networks adjacent to the last mile---enable alternate modes of control over user traffic.

   Beyond merely discriminating among packets of data upon entry into the last mile (the exclusive concern of the FCC proposals), a broadband provider can also discriminate among interconnecting networks. Comcast or Verizon could thereby "outsource" the discrimination deeper into the network by the terms and conditions under which it interconnects with these other networks. Defending only the last mile will yield an open doorway to a closed Internet.

   Compounding this shortsightedness, the relevant vertical foreclosure discussions have not been helpful. As the following discussion concerning economic analysis shows, law professors fumble outside their expertise by dressing up unsettled economic theory as if it were resolved and canonical. Their efforts, however intriguing and provocative, have avoided carefully examining the effectiveness of actual network regulations.

   Second, we detail how the FCC, like academia, erroneously translates normative standards of nondiscrimination and fairness into rules for Internet engineering. These efforts fail because any network will carry some traffic types better than others. Building and configuring a network to route packets will necessarily discriminate. An almost unconscious importing of earlier telephone or telegraph regulation into the Internet policy debate accounts in part for the mistake of blanket opposition to discrimination. (12)

   As a regulatory concept, discrimination made sense with the old, telephone circuit-switched network, which involved distinct and exclusive connections, i.e., circuits, over which people communicated. At switch points, circuits could be mandated for equal treatment because a circuit was a real and singular item. Previous "open network" regulation--such as the Modified Final Judgment's mandated equal access for competitive long distance companies or the local competition under the 1996 Telecommunications Act ("the 1996 Act")--required quite simply that networks create places at which competitors could place their wires in an equal manner. (13) Discrimination, therefore, had a clear and distinct meaning in the old telephone network.

   In packet-switched networks, like the Internet, the circuits are virtual. Legal scholars have yet to comprehend how the virtualizing of the circuit has shattered the possibility of coherent nondiscrimination regulation. (14) The crucial and difficult inquiry begins with an effort to distinguish "good" from "bad" discrimination. But any regulation must realize that no single rule can properly distinguish in a general way. Rather, the distinctions must be contextual to the consumers' experiences, markets, and social environments.

   Third, we suggest a more productive framing for defending the open Internet. The inevitability of discrimination need not paralyze policy--it merely urges a different conception of the task. To defend an open Internet does not require (to borrow the phrase from employment law) "equal, nondiscriminatory treatment" of bits or traffic as the FCC and many scholars propose. (15) Instead, it requires "equality of outcomes," i.e., equality of Internet experience, for user applications. Whether it is Voice over Internet Protocol ("VoIP"), video, web browsing, or whatever might come next, the public cares about what it can do, not the component parts of how whatever it does gets done. The obsession with packet-wise equality, i.e., a complete lack of network discrimination, is aimless. Packet-wise equality may prevent some of the worst abuses of a hostile network, but the treatment of a given packet relates only indirectly to human concern. The immediate concern is what the transmitted packets combine to accomplish in actual online experience. Contrary to the assumptions in the debate, (17) there is no necessary relationship between packet-wise equality and online experience of Internet applications.

   Determining equality of outcomes is highly contextual. It often involves issues that have little to do with traffic treatment, as we discuss in Section IV. Where others begin with vague principles, we begin with facts. We look at the prominent Internet controversies of the last decade. Our analysis shows that discriminatory treatment, as such, was not the animating concern. For instance, when Comcast inserted a packet into the TCP stream to spoof the closing of a peer-to-peer connection, this was not about discrimination. (18) We argue it was an issue of privacy and the expectation of message integrity--not a concern over discriminatory treatment. (19)

   Finally, we address the primary legal question on appeal of the Open Internet Order, which is whether the FCC has jurisdiction to impose any sort of regulation on the Internet. We argue that conventional wisdom misunderstands the impact of the 1996 Act on FCC's jurisdiction. We argue that the Act did not disrupt the existing structure of FCC regulations set forth in the Computer Inquiries and the original 1934 Communications Act: (20) the FCC regulates basic communications service while deregulating more advanced services that can be provided through the telephone network, such as FAX (21) or internet access.22 The 1996 Act's technical, and as we shall see, nonsensical definitions--"telecommunications," (23) telecommunications service," (24) and "information service" (25)--did not change this structure. Indeed, the authors of the 1996 Act would never have foreseen or desired such a result. Thus, the FCC has jurisdiction to promulgate the Open Internet Order pursuant to its authority to ensure basic communications pursuant to a regulatory structure that Congress never intended to change.

   The Article proceeds as follows. In Section II, we explore the concepts of "discrimination" and "fairness" through the lens of network engineering to properly understand how networks discriminate. In Section III, we catalog recent Internet controversies to begin growing the body of "Internet case law" and to show that nondiscrimination concerns only motivate a part, perhaps a small part, of Internet controversies. Section IV sets forth the components of a regulatory regime that responds in a bottomup manner to the complexity of Internet concerns. Finally, Section V provides an analysis of the FCC's jurisdictional authority to impose such a regime.

2. ECONOMIC ANALYSIS DOES NOT PROVIDE POLICY GUIDANCE

   The network openness debates have swirled around two economic questions. The first question is whether the last mile access provider--the ISP, such as Comcast or Verizon---can or will block...