Watching Over the Web: A Substantive Equality Regime for Broadband Applications

By Rahm, William D

This Article proposes a “substantive equality” solution for managing access to the Internet for broadband applications. It argues that the currently-proposed solutions, self-regulation and “formal equality,” either afford too much power to broadband operators or are inefficient. The model proposed in this Article accepts that the market is superior to administrative agencies when it comes to choosing which applications deserve priority on the network. Since there can be market failures, however, the Article lays out an adjudicatory enforcement mechanism to evaluate charges of harmful conduct. The adjudication would involve the application of a two-factor test for deciding which party should bear the burden of proof on the question of fairness. This method seeks to bring predictability to application developers concerned about obtaining access, while minimizing regulatory costs to network owners.

Introduction

While traveling in Tennessee on business, Doug Herring kept calling his wife but the phone calls to their home in Elberta, Alabama wouldn’t go through. The Herrings had switched to Vonage’s broadband voice service a month earlier, but they had never experienced any problems. Soon Mr. Herring discovered the culprit. His broadband provider, a unit of Madison River Communications, had blocked the Vonage application because it competed with Madison River’s phone service-a natural monopoly that today charges four times as much for a similar product.1 Unwilling to see all its profitable voice customers migrate to Vonage and unable to compete with that service’s value, Madison River simply took it off the table by denying Vonage access to its customers.2

Those who say the Internet has no gatekeeper have never heard of the Madison River case. Every time someone sends or receives an email or accesses a web page, the data transmission passes through a private company’s infrastructure. That those companies have chosen to leave the gates largely open is due partly to past regulatory policy, somewhat to economic incentives and technology, and perhaps as much to the absence of acute congestion on the Internet. As more applications compete for access, including those affiliated with network owners, the gatekeeping power of operators is enhanced. They must choose whom to favor and whom to disadvantage. These choices will influence investment and innovation throughout the economy.

Many credit the openness of the Internet for its explosive growth as a platform for economic activity and free expression.3 Unlike print, radio or television, this technology allows seemingly anyone to communicate and transact globally. In the original, narrowband world, the Federal Communications Commission (FCC or “the Commission”) reinforced this openness by imposing the existing regulatory framework for telecommunications. Specifically, “common carriage” policies prevented telephone companies from discriminating against data traveling over their pipes.4

The emergence of broadband amplified the Internet’s benefits and also increased power for network operators.5 The FCC granted broadband an exemption from common carriage requirements, essentially opening the door for operators to exert more authority as gatekeepers.6 And the Supreme Court officially upheld this FCC position in National Cable & Telecommunications Ass’n v. Brand X Internet Services? granting Chevron deference8 to the agency’s interpretation of the 1996 Telecom Act without taking a position on the wisdom of the policy.9

Today, therefore, network operators are well-positioned to exert tremendous influence as Internet gatekeepers. Technological convergence wrought by broadband, which allows previously disparate media to traverse the same network, further empowers network owners to impact multiple industries and regulatory structures.10 At least two forms of harm could develop from flawed management of the network. First, consumers could lose access to valuable services.11 Operators might advantage their own broadband services or simply protect legacy products that are threatened by broadband substitutes, even if these services are less valuable to consumers than independent offerings. second, innovation in applications may be deterred. By discriminating against certain applications, operators will discourage investment in these services.12

“Network neutrality” has become the catch phrase for a new regime proposed to regulate access for broadband applications. On both sides of the debate, big businesses have lined up to do battle. On one side, cable and telephone companies favor self-regulation, an unrestrained market solution.1 Commentators supporting this position argue that the Madison River case was an anomaly and further contend that the costs of regulation would exceed the potential for harm from self-regulation.14 They claim that the market is the best mechanism for deciding how to treat applications-whether to restrict some, grant others priority and charge for fast delivery. Moreover, they insist that any limitation on the operators’ ability to offer their own applications and manage their own network discourages further investment in broadband infrastructure, and may even violate their constitutional property and free speech rights.15

On the other side, web and software firms want to disable the gatekeeping power of broadband firms. They propose that the FCC issue rules requiring operators to treat all applications alike.16 They aspire to a system that does not distinguish among applications or attachments, similar to the electricity system or the regulatory framework of common carriage governing traditional telecommunications services.17 Such a regime, in its strongest form, would prohibit broadband firms from placing any restrictions on data traveling through the pipes.18 Advocates of this “behavioral” approach argue that innovation among applications demands harsh limits on the ability of operators to advantage their own services at the expense of others.19 Although the FCC has tacitly endorsed the premise that maintaining such access for broadband applications is important, it has stopped short of issuing rules in explicit support of this concept.20

Resolution of this debate requires consideration of two interrelated goals:21 maintaining incentives for firms to invest in network build-out and protecting innovation in applications. An inherent tension exists as applications and operators both struggle for greater power over the network. Operators want to maximize the value of their assets by controlling access to users while application developers seek unimpeded access to the same end users. But this tension is not irresolvable. New applications both generate their own value and contribute to the value of the infrastructure by acting as complements since users’ willingness to pay for broadband access is a function of the applications available to them.22 Policymakers have the opportunity to strike the right balance of encouraging development of new applications while maintaining strong incentives for infrastructure investment.

Ultimately, Congress must decide how to design such a policy. As an “information economy,” where success depends on knowledge increasingly gained from online media, the United States cannot forsake either applications or operators.23 Today, the United States has fallen behind other countries in the availability of broadband service, ranking twelfth globally.24 Pursuing universal adoption of broadband is essential not only to our continued economic growth but also to bridging social inequality.25 Democracy requires policing Internet gatekeepers as broadband becomes the predominant vehicle for educating voters.26 Currently, Congress is holding hearings and debating drafts of a bill to update the 1996 Telecommunications Act.27 It should take this opportunity to articulate how to achieve a level playing field for broadband applications.

This Article offers Congress an alternative regime, distinct from either self-regulation or network neutrality. Starting from the premise that true equality among applications requires differential treatment, this model endorses a substantive equality principle as the governing legal standard. This approach accepts that the market is better than administrative agencies at choosing which applications deserve priority. But it also acknowledges that market failures do exist and so a legal principle without an enforcement mechanism will prove inadequate. Therefore, it proposes ex post adjudication to evaluate charges of harmful conduct by broadband operators. Grounded by a two-factor test for deciding which party should bear the burden of proof, this method seeks to bring predictability to application developers concerned about obtaining access while minimizing the regulatory costs to network owners.

Part I introduces the business and legal landscape upon which any law must operate. First, it describes the industry structure. Technological and competitive forces as well as the presence of state regulatory bodies impact federal regulation in this field. Second, it highlights the key economic concepts that influence the behavior of the relevant actors. Economictheory suggests that, subject to exceptions, an operator may not have any incentive to discriminate against independent applications, even as it vertically integrates into competitive services. In addition, the ability of users to impose congestion costs on other users constitutes an externality that produces inefficiencies. Third, Part I discusses the constitutional constraints and broader legal background which will shape the choice of policy.

Part II analyzes the two popular models to manage access for broadband applications-a self-regulation approach and a network neutrality regime. The self-regulation response arises from the principle that the market, left unrestrained, will sort out questions of broadband access in the most efficient manner possible. This position attributes the absence of significant anticompetitive behavior in the past to strong economic forces preventing such conduct. The introduction of new services and technology, however, may both enable and encourage network operators to discriminate against unaffiliated content, especially if they possess market power over consumers. The significant direct costs that operators may impose through exclusion of certain applications-namely, raising the price and availability of services-as well as the indirect costs of stifled innovation suggest that self-regulation is inadequate.

Network neutrality proposals endorse a formal equality approach to access, forcing networks to treat all applications exactly alike. Various network neutrality proposals all tend to anticipate a policy that is executed through rulemaking. Not only are the regulatory costs of rulemaking in this context substantial, but the underlying principle of formal equality does not fit well with heterogeneous applications and congested networks. By prohibiting any priority among applications, network neutrality implicitly favors those services that are not affected by time delays. Thus, this regime replaces one market failure with new market distortions that may reduce incentives for investment in certain applications, as well as in broadband.

Finally, Part III develops the substantive equality regime. First, it calls for Congress to codify a substantive equality principle to govern operators’ treatment of broadband applications seeking access to the network. Since such applications are not all alike, formal equality is inadequate for achieving substantively equal treatment. Recognizing that the market generally is better suited for sorting out the appropriate treatment for a specific application, this model permits operators to choose, in the first instance, how to treat applications. To enforce the legal principle of equality, the statute instructs the FCC to issue disclosure rules forcing transparency regarding operators’ choices for differential treatment and grants jurisdiction for the Commission to hear complaints of unfair conduct. In order to provide further predictability to the resolution of such matters, the law should direct judges to employ a two-factor test to decide which party should bear the burden of proof. If the plaintiff can show that the operator has market power and did not use a fair method for choosing which applications to prioritize or disfavor, the operator will be forced to overcome a presumption of harm by demonstrating that it still satisfied the law’s substantive equality principle. If the plaintiff cannot make this showing, the operator will receive a presumption of legal conduct that must be overcome by the plaintiff in a fact-based inquiry. This method of implementing the legal standard avoids the problems associated with rulemaking while providing a mechanism for the federal government or private parties to challenge what they consider harmful treatment.

I. The Landscape Faced by Legislators

A. Internet Technology: A Multi-Layered Group of Networks

The Internet is comprised of several layers built on top of a physical platform that is itself a network of networks. As Figure 1 indicates, networks may be separated into four horizontal layers and the ability to reduce all media (data, voice and video) into a standard code enables them to all travel over the same physical infrastructure.28 Distinct entities may operate in each of these layers, but network operators have increasingly extended their presence forward from the physical layer to applications/content.29

Figure 1: Layers of Broadband Architecture30

The physical layer is itself divisible into three network components, according to the FCC. ‘Backbone providers facilitate long-distance connections between a small number of interconnection points. Middle-mile providers carry data from these interconnection points to distribution facilities. And last-mile providers convey the traffic from these facilities to end users. The networks discussed in this Article are the last-mile (or local) networks because they represent the connection point between users and all the services available through the Internet.

While some scholars have compared broadband networks to electric utilities,32 such a comparison is not as much a description of today’s situation as it is an aspiration of those seeking to limit the gatekeeping power of operators. First, electronic applications tend to be homogeneous and static in the sense that there has been little differentiation in the way electronics access the electric grid, while broadband applications are constantly developing and may require time-sensitive or standard delivery.33 This distinction alone belies any sincere attempt to analogize broadband and electricity networks. In addition, electric utilities are so heavily regulated, at both the state and federal level, that it is difficult to equate the two industries.34 Finally, electric utilities have very little information about, or control over, the last-mile of their distribution networks. Whether this resulted from design or regulation is less important than the fact that today electric utilities do not possess the same functionality as a broadband operator.

Broadband networks, by contrast, are designed to allow their owners to discriminate in the applications traveling through their pipes.35 One writer explained that “Cisco… [has] developed ‘policy- based routers’ that enable [broadband] companies to choose which content flows quickly and which flows slowly.”36 Such technology gives these broadband companies power more analogous perhaps to a cable television operator who can control its subscribers’ access and content. This analogy to cable television does not work well because of the diversity of applications competing for access on broadband networks, but it does underscore that operators are gatekeepers, a key distinction between broadband and many other networked businesses.37

B. Broadband Industry Structure: A Competition Among “Natural” Monopolies

Concentration in the market for broadband access is the most important feature of the industry structure. Concerns about anti- competitive conduct by vertically-integrated firms are heightened when those firms possess monopoly power in the primary market.38 In other words, vibrant competition in the market for high-speed local broadband access would substantially mitigate any concerns that the network owners might unfairly compete with search engines, email services, or other application-level providers. Indeed, many scholars have argued that a monopoly in the primary market is a structural precondition for harmful conduct by a vertically integrated firm in an adjacent market.39 Ultimately, if consumers have many choices and switching costs are low, there is confidence that the market will punish harmful conduct by broadband firms.40

The leading broadband technologies, cable modem service and digital subscriber line technology (DSL), emerged out of natural monopolies. A natural monopoly exists if a single firm can supply the entire market at a lower cost than two or more firms.41 Although some disagreement remains over whether cable networks and local telephone service are natural monopolies, they have long served as quintessential examples of such.42 Intuitively, the state-protected access to individual households that these firms have gained appears to provide a substantial advantage against new competition. Thus, when cable and telecommunications firms began offering broadband services, they did so from a strong competitive position relative to other technologies, for example wireless or fiber optics. And although some may claim that the field is wide open, it is difficult to deny the significant head start these products received.43

Unsurprisingly, these two technologies dominate the market for broadband, together representing 93% of all broadband subscribers nationally.44 Moreover, as extensions of natural monopolies there is no intra-technology competition, e.g., only one cable modem service provider exists in a city because only one cable company owns the local franchise. Despite the historical monopoly position enjoyed by firms offering cable broadband and DSL, some commentators still insist that the “existence of a natural monopoly does not necessarily preclude competitive entry … so long as entry and exit are easy … ’45 But the absence of new entrants into cable TV or local telephone service supports the logic that sunk costs provide a barrier to entry.

The degree of concentration depends, in part, on whether one measures at the local or national level. Recent consolidation among telecom (e.g., AT&T and BellSouth) and cable companies (e.g., Comcast and Adelphia) may render this distinction moot because we are left with only a few large players in any one community and across the nation. Still, antitrust analysis places significant weight on defining the geographic market and since the underlying platform is two-sided there is some basis for both a local and national framework. A leading advocate for a nationalperspective, Professor Christopher S. Yoo, argues that “network neutrality proposals are designed to limit the exercise of market power not [against] end users, but rather [where] last-mile providers meet ISPs and content/application providers.”46 Thus, Yoo believes a national market makes more sense because web companies, for example Google, are focused on access to customers in the aggregate, and not just in a specific region.47 On the other hand, a local definition of the market may make more sense since the harm at issue focuses on operators discriminating against what application reach end users. This matter is substantially mitigated if users can respond to unfair treatment by replacing their broadband providers, but cases like that of Madison River do not seem less problematic simply because services like Vonage can find customers in other states. Today, the FCC defines the market locally and this Article proceeds from that premise.48

Using this local perspective, the market for last-mile broadband access appears highly concentrated on average. One metric in modern competition policy is the Hirschman-Herfindahl Index (HHI). HHI is calculated by adding the square of the market share of each competitor, where a total monopoly results in 10,000 (100%^2) and infinite fragmentation yields a score of 0 (0%^2).49 The DOJ Guidelines indicate that antitrust authorities are unlikely to challenge a vertical merger unless HHI exceeds 1800 in the primary market.50 Such a score results from five to six competitors of equal size. Since local markets are typically dominated by two players- the original natural monopolies of cable and telecommunications- defining the market locally yields HHIs often above 5000.” Even using Yoo’s proposed national view, there is a high level of concentration among telecommunications and cable companies which has only increased during the last ten years.52

Many point to new technologies as likely to reshape the broadband industry.5 Although wireless, satellite and powerline technologies have seized some market share in recent years, claims of a “competitive free-for-all” seem overblown.5 Significant technological, economic and regulatory barriers limit the mass- adoption and full-scale competition of these rival services. Satellite has traditionally been a one-way transmission, i.e., broadcasting from a central point to dishes on people’s homes but not the other way. Additional technology would need to be deployed, essentially a new network, to facilitate interactive satellite communications. The most robust wireless technologies are not yet cheap enough to compete against cable and DSL. WiMax, for example, may cover a large footprint with significant capacity (e.g., 50 Mbps over a 5 mile radius) where each dish costs perhaps five thousand dollars. The problem is that providing enough capacity for each user at peak time in a populated area would require blanketing the geography with dishes which would increase the capital cost per user to prohibitive levels.55 Finally, broadband over powerlines (BPL) exhibits many of the advantages of cable and DSL-primarily, universal reach and leveraging a natural monopoly’s infrastructure- but obstacles remain. Since the electric distribution industry is considerably more fragmented than cable or local voice, BPL providers (usually independent firms) must negotiate with dozens of utilities in order to establish a sizeable footprint. Absent a federal order preempting state regulators in this field, BPL firms must presently negotiate with each utility, which substantially slows the deployment of the service.

Two non-technology based responses appear even less likely to disrupt the current concentration of broadband access. First, municipal broadband programs have been initiated by cities hoping to reap the benefits of broadband penetration among their residents. Although cable and DSL firms owe their positioning partly to state support, these firms have stridently objected to the use of tax dollars funding competition to private enterprise.56 Using their political clout in Washington, they have managed to get a Congressional bill put forward which would substantially limit such municipal broadband programs.57 Second, the cable industry’s victory in Brand X essentially relegates line sharing to extinction.58 Line sharing meant that unaffiliated ISPs would be permitted access to the last-mile infrastructure of DSL and cable providers-a regime known as “open access.”59 Since ISPs manage the flow of last-mile network traffic, a line sharing rule provides a structural remedy to discriminatory conduct by the owners of the physical networks. Consumers are granted a greater choice over their gatekeeper and thus vertically integrated operators are less likely to discriminate against certain content. Although economic and legal arguments can be made in support of affiliated ISPs and against line sharing, one must acknowledge that the demise of “open access” strengthens cable and DSL firms.60 The FCC’s support of the Brand X ruling and subsequent extension to DSL service is a rejection of “open access.”61 Although Congress could conceivably undo this result, the political power of operators makes such legislation infeasible. Without such a structural response available, policymakers seeking to limit the power of broadband firms must consider a conduct approach.62

In reviewing the industry structure, it is worth noting that other layers of the Internet appear significantly more competitive. While some large scale firms exist in applications, e.g., eBay in auctions, their power is still of a different order than that of the platform provider. Not only has no upstream player attempted to integrate backward into the physical layer, but they also do not derive their power from an underlying natural monopoly.63 Thus, broadband service providers integrating forward confront a highly fragmented market in sharp contrast to the concentration of their primary market.

Although the foregoing illustrates the broadband industry today, its nascent stage of development urges caution against views that it will always look the same. New technologies still offer the potential of competing platforms, especially if BPL receives some expediting at the state level (or through preemption) and if wireless overcomes its economic limitations. For the foreseeable future, however, it appears that most users will have little or no choice among broadband operators. The introduction of switching costs only aggravates the market power of incumbents. Thus, the primary market appears sufficiently concentrated to inquire into the role of private firms as gatekeepers of the Internet, especially as they integrate forward to compete with other applications.

C. Economic Concepts: Congested Networks in a Vertically Integrated Industry

1. Network Effects

A networked business provides a platform and manages interaction among members.64 A “network effect” arises when one user’s value of participating in a network increases with the addition of more users to the network. The fax machine serves as the paradigmatic case for such a phenomenon. A fax was useless to its first purchaser until a second purchaser enabled the two to communicate. As more users joined, the utility of the fax for the first user increased because she could communicate with more people. This initial user paid nothing to additional users even though their membership created incremental value for her.65 Thus, additional participation yielded a positive externality benefiting the first user.

The two most common network structures are homogenous, as seen in most communications technologies, and heterogeneous (hereinafter, “a twosided network”).66 Homogenous structures are simply those in which users exchange roles frequently-the sender of the fax is often the receiver. Economists offer a more technical definition: “[T]he market for interactions between the two sides is one-sided if the volume … of transactions realized on the platform depends only on the aggregate price level.”67 In a homogeneous structure, reallocation of the price between buyers and sellers will not change the volume because buyers and sellers operate in both roles repeatedly.

Two-sided networks, by contrast, have distinct sets of users who consistently play their defined role in transactions. Success with such a network requires not only choosing the right pricing level, but also the appropriate pricing structure since differential pricing of each side can meaningfully impact aggregate volume. For example, video game businesses have a platform (usually a console like Xbox), end users, and game developers. The platform provider must decide what to charge end users for the console and whether to charge application developers for access to the codes that allow games to be produced for that system. The developers will have a higher demand for the platform if there are many end users and the end users will enjoy more value from their participation as the number of games available increases. Thus, platform owners have to solve a “chicken and egg” problem: how much to charge each side in order to seed enough users to create value for the other side, initiating the virtuous circle.68

Broadband technology exhibits two network effects because it is both a standard communications infrastructure (“homogenous”) and a platform similar to the video game company with end users and application developers (“heterogeneous” or “two-sided”). As a pure communications vehicle, all users of the network are homogeneous- both senders and recipients of information-who gain value from having more participants on the network with whom to communicate. From another perspective, broadband is a twosided network because applications and content are complements to technologies like cable modem service and DSL. Companies like Google, Vonage and Amazon will develop more s\ervices when they are confident that there are more users whom they can reach. On the other side, the presence of these applications increases users’ willingness-to-pay for broadband and therefore fuels user adoption. Promoting both innovation at the applicationlevel and adoption by users requires managing the network to attract both sides.

Platform owners are in the best position to capture the value of network effects. As the willingness to pay increases among both users and developers, broadband firms can charge each side more. Currently, one side-the developers-do not really pay for access to the platform. The absence of any pricing on this side may result in overconsumption that can produce negative externalities. In addition, the ability of platform owners to integrate forward and compete with applications adds complexity to the use of this network effects theory.

2. Economics of Congestion

Frequently referred to as an “information superhighway,” the Internet is subject to congestion much like physical highways. When networks are subject to congestion, negative externalities arise because one customer’s usage can degrade the quality of another user’s service.69 Imagine, for example, an airport security line. If a traveler shows up and there is a free station, she walks through. If, instead, the traveler arrives and there are other passengers ahead of her in line, the traveler will feel a cost of that delay. In some cases, the traveler is time-sensitive because delay could cause her to miss her flight. Indeed, she might be willing to pay a fee in order to jump ahead in line. If a mechanism existed to identify relative demand among those passengers waiting in line, the security personnel could attempt to re-order the line based on demand rather than on its current system of first-come, first-serve in order to maximize welfare among travelers.

Congestion exists on the Internet, too. Just like passengers entering airport security, when too many data packets arrive at the last-mile network, they form a queue. The resulting delay in speed reduces the quality of service.70 A major cause of this congestion is over-consumption by a small set of users. Typically, broadband customers pay a flat-rate monthly fee for unlimited service. The pricing is established in part based on the return that operators demand on the amount of infrastructure they project will be required by the estimated usage of those customers. Statistics confirm that while end users pay a flat-rate monthly fee for service, consumption is not distributed evenly. BellSouth, for example, claims that 1% of users drive 40% of Internet traffic. Since many facilities are shared architectures, i.e., dozens of consumers may access the Internet through one pipe, “over-consumers” will significantly slow other consumers’ service, as operators do not deploy sufficient capacity to meet demand.72

Not only is congestion produced by a disproportionate few, but the costs of this externality are not uniformly distributed. The airport security example helps illustrate how certain users feel the effects of congestion most acutely. Missing her flight was a much greater cost to the late traveler than waiting for an additional person to clear security would have been to the passengers in front of her. In the broadband context, the costs of congestion for time- sensitive applications, such as streaming media or voice over internet protocol (VoIP), are far more severe than the impact to other services, like email. Indeed, VoIP or video products may fail to perform with too much delay, equivalent to missing the flight altogether, while email will simply arrive in the inbox a moment later, where it may languish for hours just like the passengers who have arrived early for their flight. The essential differences in the character of applications would be irrelevant if capacity were infinite, eliminating all congestion. But just as there will always be delays at some security checkpoints no matter how many additional ones are added, so too is there likely to be significant congestion on broadband networks for the foreseeable future.73

Two primary responses exist for congestion. The first is usage- based pricing, which is intuitively appealing because the Internet is a club good, i.e., a good shared by more than one person.74 Reliance on flat-rate pricing often results in inefficiently high levels of congestion and over-consumption of broadband.75 Original access to the Internet, through dial-up, was accomplished through a “pay as you go” model. The economic intuition is simply that the private cost of consuming an additional unit of capacity is zero under a fixedrate regime so utility-maximizing users will increase consumption until their marginal utility is zero.76 In the process, they will create substantial congestion costs for others.

Despite the appeal of usage-based pricing, fixed-rate regimes dominate broadband.77 Many argue that this model contributed to the explosion of Internet usage as consumers could “web surf without incurring incremental charges.78 Today, consumers expect “always on” and unlimited fixed-rate service and changing that expectation may be difficult. In addition, metering is required for usage-based pricing and that requires significant transaction costs likely to deter broadband firms. Indeed, every communication on the Internet is broken into smaller packets that are transmitted individually and reassembled at their destination. Thus, each communication requires multiple records. One study estimates that a ten-minute phone call over the Internet would require tens of thousands of records to account for the associated packets.79 While new equipment may enable broadband providers to dramatically lower this metering cost, it does not appear that usage-based pricing will become economical in the near term.80

Instead of usage-based pricing, network owners might mitigate congestion costs by restricting certain applications and prioritizing others. At the extreme, this approach could take the form of blocking certain applications or practices such as unauthorized re-selling of capacity, home networking, attachments, or commercial uses.81 It is important to see how these uses contribute to substantial congestion. For example, bandwidth resales impose congestion by transforming a single connection into one serving multiple end users.82 The corollary, returning to the airport security example, would be one ticketholder bringing an entire group with her through the security line. The other members of the group are not contributing whatever fractional share of a ticket goes to supporting the security line, but they are imposing a cost on other users. Similarly, when someone buys access through their home but shares it with multiple other users, they are paying a fee far less than the demand they are imposing on the network. Prohibiting this activity might preserve fixed-rate pricing while mitigating the externality of congestion by freeing up more capacity.

Alternatively, network owners might become even more sophisticated managers of applications in order to limit the effects of congestion. As previously mentioned, applications pay no fees beyond the typical access charges. Instead of monitoring end users’ consumption, network owners could simply require those applications that require time-sensitive delivery to pay for access.83 The platform would not be forced to meter usage but could use demand from applications as a proxy-an application provider would presumably only pay for time-sensitive delivery if users cared about avoiding delay. Under this system, the time-sensitive application could pay for priority which allows it to perform properly and gives the operator more funds to invest in further capacity. An equivalent scenario would be charging the late traveler in exchange for permitting him to skip to the front of the line. Over time, the airport security office could afford to provide an extra station for business- and first-class travelers who valued avoiding delays, and in the process they would relieve some of the congestion for other passengers too.

Lawmakers considering a regime for managing broadband access must acknowledge the negative externality associated with congestion. Sophisticated technology might enable usage-based pricing to become a cost-effective solution to this problem. In the absence of such a metering response, network operators have an interest in treating applications differently in order to maximize welfare among applications and, ultimately, end users.

3. Vertical Relations

Any framework for governing access to broadband applications must not only anticipate the value of network effects and the costs of congestion, but also the implications associated with network owners creating “vertical relations.”84 These relations include not just integration into applications through merger or organic growth, but also contractual agreements with firms in related markets. The 2001 Southwestern Bell Company (SBC) and Yahoo! pact offers an example of a vertical relation. SBC provides DSL service in over a dozen states. After the SBC-Yahoo! agreement SBC’s users received a subscription to Yahoo!, which made Yahoo! the default homepage for all SBC DSL users. Although SBC did not limit its users from accessing content by Yahoo!’s competitors, such an exclusive relationship was not prevented by law. Moreover, Yahoo! might have contracted to receive prioritized delivery on SBC’s network such that its data would leap ahead of other competitors’ data if congestion required queuing.

Telecommunications regulation and antitrust policy traditionally have diverged in their treatment of vertical relations.85 In the 1970s, the Chicago School influenced mainstream antitrust thinking to accept that vertical relations could provide efficiency benefits to consumers. In the context of a complements-based business, a s\trict separation of the application providers and platform owners can lead to costly hold-up hazards or free riding by application firms.86 Moreover, vertical integration can eliminate double marginalization where two monopolists (one in the application layer and the other in the physical layer) both impose a monopoly markup.87 Finally, as Professors clayton Christensen and Michael Raynor have extensively argued, if the interfaces between applications and a platform are not well defined, innovation can be slower and more costly, and integration may be preferable.88 As a result, courts dealing with antitrust claims often presume such vertical relations are unobjectionable unless a fact-intensive inquiry shows otherwise.89

By contrast, starting in the 1970s, the vertically integrated AT&T network was broken up as a result of a policy to develop and protect open interfaces.90 The philosophy underlying the breakup held that “powerful firms at one level should not be allowed to leverage that power into-or perhaps even participate in-adjacent competitive segments.”91 The 1996 Telecom Act initiated a new regime that promoted competition among local exchange carriers (e.g., the “Baby Bells”) and long-distance providers (e.g., AT&T, MCI, and Sprint) for customers, allowing each to integrate and compete in the others’ traditional spheres of influence. Recently, mergers of SBC/ AT&T, and potentially BellSouth, as well as Verizon/MCI may mark the full-scale reintegration of different network verticals.

Analysis of vertical relations requires evaluating the competitive position of firms at each level.92 If there is robust competition at the physical layer, concerns about integration are muted because consumers can always switch among broadband providers.93 As previously discussed, the last-mile broadband market is currently characterized by two dominant natural monopolies present in most, if not all, markets. Cable and DSL may compete against each other, but their statuses are defined by their origins from technologies that had almost exclusive access into residential homes. Such access is prohibitively costly to replicate, thus relatively immune from competition.

Different theories address the likelihood that a monopolist broadband provider will inefficiently discriminate against unaffiliated applications. The “one monopoly rent theory” claims that a monopolist has no incentive to monopolize a complementary product if it is used in fixed proportions with the monopoly good and is competitively supplied.94 Joseph Farrell and Philip Weiser introduce the concept of internalizing complementary efficiencies to argue that a network owner will only deny access to unaffiliated competitors if it is efficient to do so.95 It is unsurprising that where a platform provider chooses to focus on maximizing the wealth of its platform, and avoids forward integration, it will seek to maximize competition and innovation among applications so it can leverage the network effect that motivates user demand.96

Where a platform monopolist does integrate into the applications market, Farrell and Weiser’s theory claims it will still “welcome value-added innovations by independent firms … in order to profit from a more valuable platform.”97 The argument assumes that providers can extract more value from increased demand for the platform than from their gains through sales of applications. Importantly, the theory allows for some limitations on competition in the applications market which benefit the platform’s users. Restricting some applications to manage congestion is one conceivable example where operators might attempt to limit access for applications in order to maximize welfare for users.

These limitations, however, may result in an inefficient outcome for users when the platform monopolist can gain more from exercising market power in applications than it can from increasing demand for the platform. The classic example occurs where pricing for the platform is regulated, perhaps below the profit-maximizing level, and thus the network owner seeks to monopolize the applications market in order to take additional profits in that market, perhaps inefficiently.98 This reasoning supports unregulated pricing of broadband access to prevent operators from acting unfairly in the applications market to compensate for platform pricing set below the profit-maximizing level. Another more relevant exception occurs when operators maintain separate legacy businesses threatened by broadband applications. Specifically, VoIP and IPTV serve as substitutes, and thus competitive threats, to traditional voice and cable TV frequently offered by the same firms who sell broadband. The Madison River case, described at the beginning of this Article, offers the paradigmatic example of a phone company restricting access for a broadband application in order to maintain its more expensive traditional service at the expense of its broadband users. Even if Madison River had offered a competitive VoIP service, it would likely have slowed Vonage in order to maintain its subscriber base of voice customers through a broadband application if its legacy product were no longer viable.

Lawmakers, therefore, must acknowledge the tension that exists in these vertically integrated broadband providers. First, due to network effects, a platform provider has strong incentives to maintain competition in its adjacent markets because robust innovation in applications increases the platform’s value. Broadband firms can appropriate some of this increased value in the absence of price regulation. Furthermore, restrictions are not per se inefficient because certain services may impose such severe harm on other users (e.g., viruses) that operators should be encouraged to limit their access.

On the other hand, broadband providers may seek to engage in exclusionary conduct that harms consumers.” There are many subtle ways to disadvantage an unaffiliated developer (e.g., an interface design that slows the application or an onerous pricing policy) that can benefit the network owner.100 An operator may seek to increase its power in the applications market over competitive services by restricting or degrading the independent developer’s access to its users. In such situations, a network owner can weaken a rival by shrinking the pool of users that the rival can reach.101 The prospect of inefficient discrimination by broadband firms requires lawmakers to construct a model that evaluates both the market power of the firm in its primary market and the nature of its differential treatment of applications.

D. Legal Background

Developing effective policy in the arena of broadband access requires close attention to several legal principles since the Constitution may limit certain regulatory approaches and since existing legal doctrines may already have some influence over the issue. Sharp conflicts emerge in discussion of the Constitution’s position on access. Network owners argue that regulation amounts to an unlawful taking in violation of the Fifth Amendment. Conversely, concerns about broadband firms restricting access to undesirable applications or even specific viewpoints has motivated First Amendment arguments in support of government intervention-although the speaking rights of the network owners themselves may pose free speech claims cutting against regulation. The complexity of regulating in this area is amplified by the presence of both federal and state commissions which may require some sensitivity to the role of the federal administrative state. And finally, any model will operate in the shadow of antitrust jurisprudence that already seeks to limit unfair use of market power by dominant industries. While exploring these broad fields of law in detail is beyond the scope of this Article, it is possible to highlight the major issues and the analytical frameworks that lawmakers should consider.

1. Fifth Amendment

Many advocates of a deregulatory approach to broadband access have sought to ground the discussion as a property rights debate.102 Indeed, they have even claimed that supporters of network neutrality are engaged in a “crusade against property rights in broadband networks.”103 Although it is undeniable that operators have a real property interest in their infrastructure, most proposals for prohibiting discrimination against unaffiliated content do not raise any significant Fifth Amendment concerns.

The Takings Clause analysis depends significantly on whether the regulatory approach includes a physical invasion.104 Daniel Spulber and Christopher Yoo make this point effectively when they note that “[Regulation [that] simply adjusts the terms under which parties can contract… is subject traditionally to a rather permissive standard of review under the Takings Clause …. Compelled] access to a physical network, in contrast,… [is] subject to the more restrictive standards associated with the Court’s physical takings jurisprudence.”105 The structural approach of open access-requiring broadband firms to allow the equipment of unaffiliated ISPs on their linesmay entail such a physical invasion, but regulations against discriminatory behavior affect only the “terms” under which operators contract with application developers and end users.106

Regulations affecting terms of use rarely constitute a taking. In Lingle v. Chevron, the Supreme Court recently clarified how to evaluate a takings challenge.107 Writing for a unanimous court, Justice O’Connor limited per se takings to two categories: (1) where the government requires an owner to suffer a permanent physical invasion of her property108 or (2) where a regulation completely deprives an owner of “all economically beneficial use” of her property.109 Since non-discrimination in broadband access does not contemplate any restriction on charging end users, such a regulation does not deprive network owners of “all e\conomically beneficial use” of their property.110

Outside of these two per se categories, the “ad hoc” test laid out in Penn Central Transportation Co. v. City of New York governs regulatory takings challenges.111 The test includes (1) the character of government action, (2) the severity of the economic impact, and (3) the degree of interference with reasonable, investment-backed expectations.112 An inquiry into the purposes of broadband access regulation shows that a challenge here is likely to fail. First, such a policy intends to increase demand for broadband among both consumers (who will have more content) and application developers (providing them the certainty of reaching users). By increasing demand for the platform, the regulation increases its value and, subsequently, the economic return for the platform owner. Even if operators might improve their return more through discrimination, regulation restricting such activity hardly has a “severe” economic impact and essentially amounts to an adjustment of the “benefits and burdens of economic life for the common good.”113 The Fifth Amendment, therefore, provides little assistance to critics of non-discrimination policies for access to broadband applications.

2. First Amendment

By contrast, proponents of regulation cite constitutional concerns with self-regulation. If broadband networks become the primary means of communication, they ask, does the power of operators to exclude objectionable content threaten freedom of speech? At an abstract level, the First Amendment “focuses on the importance of securing an open environment in which all can equally experiment with how to think and speak, and where no one can determine for anyone else what is orthodox.”114 Some have even argued that “the First Amendment embodies an affirmative right of access to the means of speech, judicially enforceable against a government agency that does not adopt policies that secure access to the means of effective communication.”115 The courts have not gone this far, but the First Amendment still supplies a basis for regulating fair access to broadband content.

Two separate strains of constitutional jurisprudence may support regulation in this area. First, in Pruneyard, the Supreme Court held that the free speech rights of a shopping center owner were not violated by an order to allow patrons of his center to peaceably express their views, even when he disliked them, so long as he continued to open his center to the public and maintained the ability to disclaim those views not his own.116 Since the broadband network is also private property opened to the public for commercial and expressive purposes, perhaps operator restrictions on objectionable content constitute a violation of the application developer’s rights of free expression. Indeed, a novel argument might be made that the state’s granting of licenses for broadband firms to market their services constitutes state action and, as a result, any burdens placed by operators on users’ rights to free expression may be imputed to the state.117

Second, the Court has confirmed the power of Congress to force cable network operators to carry certain television content it deemed importantprimarily broadcast channels. Although the Turner cases represent an acknowledgment of the free speech rights of operators, the holding established that Congress may limit those rights so long as it can demonstrate important governmental interests and not substantially burden more speech than necessary to further those interests.118

Although the First Amendment is probably better employed in support of regulation, there are some reasons to believe operators, might win a contest over speech rights. Most notably, Professor Philip Weiser points out that the FCC’s classification of broadband as an “information service,” upheld by Brand X, infuses operators with speaking rights that they would not have as common carriers.” Furthermore, a federal district court struck down a linesharing requirement for ISPs intended to limit operators control over content on the basis that the requirement violated the speaking rights of the operator, who was distinguishable from a common carrier telephone service provider.120 Although Pruneyard and Turner II still cast doubt on the use of the First Amendment against broadband access regulations, some advocates would like to see the FCC clarify the issue by expressly disclaiming operators’ speaking rights.121

3. Federalism

Communications have long been overseen by a patchwork of government regulators. Since the establishment of the FCC under the Communications Act of 1934, the federal government has been authorized to regulate interstate and international communications by radio, television, wire, satellite and cable.122 State and municipal bodies, however, retained an active role through granting cable franchises and setting rates.

Leaving the issue of overseeing broadband access to states is illogical and out of step with recent congressional action. The Telecommunications Act of 1996, for example, sought to promote competition and reduce regulation to “encourage the rapid deployment of new telecommunications technologies.”123 A uniform approach is necessary to achieve this goal, especially given that the footprints of broadband firms frequently exceed state borders. Courts have already demonstrated a willingness to support federal preemption on this issue, but lawmakers should consider more comprehensive legislation in this area.124

Brand X concluded that the FCC is permitted to classify broadband transmission as an “information service,” but also ruled on the full extent of the FCC’s authority to regulate such service. Philip Weiser proposes that the FCC could be entitled to oversee the Internet “by reference to its need to regulate ‘information platforms’ that can support the delivery of voice, video and text applications.”125 As he points out, in today’s world, broadband access itself is the platform upon which providers will rely to offer their Internet-delivered content and services. Thus, the content that was previously delivered directly by a physical coaxial cable or copper wire is now transmitted through broadband networks. As this platform replaces the legacy infrastructures, the FCC can only achieve its express statutory purpose if it is empowered to oversee the management of those networks.126 Either the FCC can employ this regulatory model and seek congressional endorsement later or Congress can affirmatively extend the agency such power in a new bill-since the 1996 Telecom Act only barely anticipated the rise of broadband.127

4. Antitrust

Finally, antitrust doctrine, with its purpose of constraining unfair use of market power, serves as a backstop to any regulation in this area. Two cases inform the antitrust analysis of access for broadband applications. The Supreme Court’s landmark decision in Continental T. V. v. GTE Sylvania, Inc. replaced per se prohibitions of vertical restrictions with an evaluation under the “rule of reason,” which entails a fact-intensive inquiry into the competitive effects of the restriction.128 Although the dispute in Continental T.V. dealt with the relationship between a manufacturer and its franchise retailers, Justice Powell’s majority opinion embraced the idea that “[v]ertical restrictions promote interbrand competition by allowing the manufacturer to achieve certain efficiencies in the distribution of his products.”129 Antitrust doctrine continues to presumptively permit vertical relations and thus courts may not take issue with the extension of network owners into the application and content layers of broadband.

On the other hand, where the primary market is highly concentrated, Continental T.V. may be less applicable because there is no interbrand competition to encourage. In such situations, the FCC’s Carterfone decision may offer a better analogy.130 That decision responded to AT&T’s prohibition on competitive attachments to its lines-for example, independent makers of phones-and spawned a regulatory regime based on “open access” to the telecommunications network.131 The FCC also adopted pro-competitive policies for services in markets adjacent to the incumbent telecommunications network, known as “information services.” The intention of these FCC policies was to prevent a monopolist in the primary infrastructure market from leveraging its market power to disadvantage unaffiliated competitors in the secondary services market. While Brand X upheld the FCC designation of broadband as an “information service,” it did not displace the regulatory construct from which such a designation was bora. In other words, it did not upset the idea that vertical restrictions imposed by a monopolistic platform provider must be regulated.132 Although the open access paradigm first established by Carterfone would not require a network neutrality regime for governing access to broadband applications, it does encourage some regulation to address the potential market failures associated with vertical restrictions. Philip Weiser supports an “antitrust-like approach to regulation” of services such as broadband.133 Even if lawmakers do not utilize antitrust doctrine in the area of broadband access as explicitly as Weiser suggests, any solution must consider the existing frameworks for addressing vertical restrictions by firms possessing significant primary market power.134

II. Existing Proposals for Governing Access: Self-Regulation or Network Neutrality

A. Self-Regulation: “Keep Your Hands Off the Market! “

1. Summary of the Self-Regulation Approach

Although a market-oriented solution may mitigate congestion by allowing an operator to prioritize applications efficiently, the presence of concentrated players and the potential for market failure warrants government involvement. The absence of any restrictions may prove too tempting for these gatekeepers, andeven limited discrimination may reduce social welfare by chilling innovation.135 While broadband operators have a real property interest in their infrastructure, this interest should not deter lawmakers. There is no strong legal argument against a legal principle governing operators’ treatment of broadband applications.

2. Incentives for Efficient Self-Regulation

Broadband operators assure policymakers that the absence of significant harm demonstrates the market’s ability to regulate their conduct.136 And many commentators agree, highlighting the economic forces that motivate network owners to promote competition among applications.137 They warn that regulation will not only impose significant costs, but will also deter future investment; therefore Congress should codify the customary “hands off approach that has so far been customary in order to minimize costs and maximize economic development.138 While this unrestrained market approach has its virtues, it overlooks the opportunities and incentives for vertically integrated providers to leverage their gatekeeping position unfairly at the expense of social welfare.

A key premise behind self-regulation is that competition in the primary market for broadband is vibrant.139 Government interference is unnecessary, proponents of self-regulation contend, because consumers can switch platform providers if applications are banned or degraded. Moreover, restrictions on vertical relations would not only prevent operators from earning an adequate return on the capital invested in infrastructure, but would also cut against legal doctrine, which presumes integration into an adjacent market to be reasonable as long as the primary market is competitive. As this Article has discussed, this picture of broadband competition defies the reality in most communities where no more than two operators are likely to serve end users. While the degree of concentration depends on the geographic market and perhaps more on one’s market definition, one can hardly refer to the market for broadband as a “competitive free-for-all.”140

In response to claims of concentration, advocates of a “hands off’ approach point to Farrell and Weiser’s economic theory-namely the theory that even a platform monopolist will make efficient choices about whether to impose restrictions on applications.141 In a Cato Institute release, Adam Thierer writes:

[E]ven if current [operators] have significant market power, they still have a strong incentive to carry more content and websites to maximize consumer utility and get consumers to spend more money for access to the service. If a carrier attempted to greatly curtail or limit certain types of Web services, it might discourage subscribership and thus reduce profits.142

There are even instances, as pointed out above, where an operator may justifiably disfavor certain applications to benefit users. The least controversial form of this conduct is where a network manager blocks viruses which harm users but are most cheaply addressed by the operator.143 More contentious is when certain applications, such as email, are slowed to allow other time-sensitive ones, such as voice, to get ahead in line.

Advocates of self-regulation argue not only that applications should be prioritized based on their demand for delivery, but also that the market is the best judge of which applications should receive preference.144 Although the market is most likely the best vehicle for identifying and achieving an efficient ordering of applications, it does not follow that operators should possess unrestrained discretion to make these choices. Granting operators the right to set the terms and pricing of access, including how to allocate priority positions, empowers them to engage in a host of self-serving and ultimately inefficient behavior. Some mechanism restraining operators’ decisions must exist to prevent both the direct harm of inefficient restrictions placed on valuable applications and the secondary harm of discouraging innovation through the unpredictable choices of network operators.145

Returning to the story of Madison River Communications may illustrate the potential for market failure. In mid-2006, GulfTel, a division of Madison River, offered unlimited local and long distance phone service for over $115/month. Vonage charged $25/month and many of GulfTel’s DSL customers, who paid $42 each month at the time, could have easily switched to that offering. Assuming that the quality of service was equivalent, customers should have been expected to switch because they would have saved over $90/month. Although GulfTel’s DSL service was more valuable to its users because of the Vonage offering, one can imagine that GulfTel might have raised its DSL price only incrementally, perhaps to $45/ month.146 By blocking access to Vonage, or even more subtly degrading its service quality, GulfTel may have acted rationally, but the result would have been costly for consumers. It is important to note that similarly perverse incentives may have existed if GulfTel integrated forward to offer its own broadband voice service. Although its service might not have been superior to Vonage’s product, it could have offered it for $50/month and misled consumers to believe its product as better by slowing the delivery of Vonage data or granting priority only to its own service in moments of congestion. In either case, customers may have believed that they had received a deal-saving $65/month on voice and getting a better product than Vonage. In reality, consumers would have overpaid while GulfTel benefited both from a more valuable broadband platform, which included the voice application, and from mitigating the lost revenue from its legacy service. In addition, application developers like Vonage would have been less likely to design products in the future because of the risk that broadband operators would unfairly compete with their own services.147

3. Real Costs v. Absent Harms

Beyond the incentives for efficient self-regulation, a frequent point made by opponents of access regulation is that regulatory costs are real while the