Berkman Center::Harvard Law School::Internet Law Program::July 1-5 2002

Access to the Internet

Yochai Benkler

Last updated May 31, 2002

Table of Contents

Introduction
Access to Physical Layer Infrastructure (1) - Copper Wires
Access to Physical Layer Infrastructure (2) - Coaxial Cable
License-Free Wireless Spectrum
Access to the Logical Layer: ‘Soft’ Infrastructure
Universal Service
Discussion
Additional Materials

Using the Internet depends, in the first instance, on access to the network. The initial emergence of “the Internet” in the early 1990s—from the increasing connectivity of a series of university and government networks alongside private services like America Online, Prodigy, and Compuserve, occurred almost entirely across slow dial-up modem connections over telephone wires. Sufficient for email, Usenet news groups and transferring relatively small files, and later viewing simple web pages, slow transfer made consumption of data rich content infuriating and its provision unprofitable. There was, however, an important compatibility between the Internet architecture and the plain old telephone system. The basic protocols of the Internet (click here for a tutorial on TCP/IP) treat all information as equal. They do not recognize rich content or poor content, content owned by one person or another. So too, the basic telephone network, because it is regulated as a common carrier by the FCC, was required to treat all these data calls alike. The combination meant that in this new medium, unlike in the mass media of the 20th century—television, cable, and newspapers—no one had much of an advantage over anyone else in communicating their views to the world. The low bandwidth available also meant that “production value”—expensive sets and cameras—that also limited access to the opportunities to speak in traditional mass media, were less important. The result was a substantially more egalitarian communications medium than any that the 19th and 20th century had known. For a while, and for limited communications applications.

In the mid-1990s, as the number of users grew and the graphical user interfaces to the World Wide Web emerged, the variety and complexity of the applications that could be used over the Internet increased dramatically. While modem speeds too increased, there were serious limitations on what traditional phone lines with traditional modems could do. These constraints and the prediction that the Internet would make the market in ‘high speed’ broadband access (synchronous transfer speeds in excess of 200 kilobits per second) immensely valuable led to substantial investments and something of a “race” to capture the broadband market. The vision for what was then called the National Information Infrastructure (NII) was that an entirely new network, eventually all in fiber optic, would one day emerge for high speed communications. In the long haul network this was already happening, with redundant fiber optic systems already in place by AT&T, MCI, and Sprint, as well as others. The problem was how to fiber the last mile, and the U.S. strategy in the first half of the 1990s was that the two main players—the telephone companies and the cable companies - would enter into competition in all services with each other, which would eventually result in two high capacity fibers being connected to every home.

This vision for the emergence of a NII was an important driving force, or at least an important justification offered for, the passage of the Telecommunications Act of 1996. The idea was to allow incumbent local exchange carriers to enter into the long distance and video delivery markets, to allow the cable carriers into the telephone market, and to allow the long distance carriers into both. In this competition of all players for all services, prices would decline, and redundant ubiquitous high-speed infrastructure, it was thought, would be built.

As things progressed over the course of the later 1990s, however, it became clear that most high speed access services would be delivered initially through a variety of innovations in how legacy infrastructure—the copper wires of the telephone system and the coaxial cable of the cable companies—was used. Competition increased in some areas of telephone service, but local telephone and cable markets seemed resistant to substantial competition. In particular this seemed to be true in competition for high speed Internet connections to residences. Owners of existing conduits to the home have a clear advantage but their networks require adaptation. Working on the assumption that whoever controlled the physical network would have a head-start in the 'battle for eyeballs', consolidation followed: AT&T sought to enter the local loop by expanding its cable network through the acquisition of TCI and Media One. AOL merged with Time Warner. The local exchange carriers merged, and the seven Regional Bell Operating Companies and largest non Bell local exchange carriers—GTE and SNET—combined since then into four—Verison, SBC, BellSouth, and US West/Qwest.

In the most recent FCC report on high-speed access, the Commission estimated that there are currently 9.6 million subscribers to high-speed data services, of which 5.2 million, roughly 54%, subscribe through a cable service. Only roughly half of that, 2.7 million or roughly 28%, subscribe to DSL. Of these, 93% of subscriptions are through incumbent LECs. The remainder of the lines are almost entirely business and university oriented T1 and better lines. Satellite and fixed licensed wireless put together, the two other sources of high speed connectivity often cited, account for 0.2 million lines, or roughly 2%.

Access to the legacy infrastructure has been pervaded by a concern with attempts by incumbents to use their position to gain control over the market for high-speed Internet access, whether over telephone and cable facilities. The Telecommunications Act of 1996 itself required invasive regulation of the incumbent local telephone companies’ relationships with their competitors in order to assure reasonably equal access to competitors. Its implementation has been hard fought and difficult. As for access to the cable infrastructure, the need for regulation has been a central issue of public policy debate in 2000-2001. Section 706 of the 1996 Act (see notes to 47 U.S.C. § 157) directs the FCC to ensure the ‘reasonable and timely’ provision of advanced telecommunications capability and to take steps to accelerate its deployment if necessary. In its First Report under this section, in February 1999, the FCC concluded that no regulatory action was warranted with respect to broadband, finding its rate of deployment adequate. A Second Report, based upon more extensive data, was issued in August 2000. One month later the FCC released a Notice of Inquiry Concerning High-Speed Access to the Internet over Cable and other Facilities, leaving open the question as to whether access to cable infrastructure would be mandated.

In the early part of 2002 the Commission initiated a major push towards resolving the regulatory questions it faces. It issued a Third Report on deployment in February, 2002. More importantly, it issued a series of proposed rulemakings regarding cable, incumbent telephone carriers, and in particular the unbundling requirements placed on them. It also is attempting to develop an appropriate general framework towards wireline-based broadband. These seem to be tending towards skepticism with respect to open access regulations, and an increased reliance on competition between cable and telephone, now increasingly being called "intermodal competition," rather than competition among the incumbent platform owners and unaffiliated ISPs on both the telephone platform and the cable platform, now usually termed "intramodal competition."

The difficulties raised by providing open infrastructure over legacy infrastructure raise the question of whether there are alternatives. Perhaps the most promising avenue to circumvent the problem of the proprietary last mile has been the emergence of the possibility of license-free wireless infrastructure development, which is explored in the third segment of the module. Another, but as yet undeveloped approach is the deployment of publicly owned and operated infrastructure at the municipal level, perhaps by using the sewage systems as the primary conduit.

Finally, even if the physical layer of the infrastructure for accessing the Internet is not controlled by any single entity, there remains the problem of the logical layer of the infrastructure. Just as at the physical layer, if any single company controls the logical layer--the browser or operating system, the ISP software or the messaging platform--it can exert tremendous control over the way in which that infrastructure is controlled. This problem of access to the “soft” infrastructure is explored in the fourth segment of the module.

The Open Access debate unites three distinct concerns. The first is to neutralize the anti-competitive risks constituted by concentrated control of key facilities. This aspect will be the primary focus of the discussion of the telecommunications sector policy of unbundling. The second centers on the architecture of the network and the desire to maintain its end to end (papers) architecture as the structure best endowed to support innovation. Finally, there is a wish to maintain and nurture a conduit of communication uniquely democratic in the modern era, affording comparable communicative power to commercial and individual speech, amateur and professional. Regulatory choices adjusting the market environment and future technological evolution will ultimately determine whether the communications paradigm remains mired in the broadcast era, where a few speakers dominate content provision and the mass merely consume, or whether a richer system may emerge, capable of providing greater diversity of voices and individual expressive autonomy. The choices determining the outcome of these challenges are occurring at three layers in the information access process: the physical layer (computers, communications paths, routers); the logical layer (operating systems and software), and the content layer (informational and cultural inputs). As information is a composite good, control over any of the three layers of production threatens extension of control to the whole. (If you are interested in how the physical, logical, and content layers of the infrastructure interact, you can explore here.) Much that has been covered in the other modules concerns the content layer. The following segments of this module address the contested control of the first two layers.

Throughout most of the 20th century the Bell System provided telephone service in the United States as a regulated monopoly. As competition for elements of the telephone service—primarily customer equipment and long distance—began to emerge in the 1960s, a fifteen-year process of antitrust enforcement began. This culminated in the breakup of AT&T in 1984 under what has come to be known as the Modified Final Judgment, or the MFJ—the order that settled the AT&T antitrust case. After the breakup, local telephone service continued to be a monopoly, with the seven Regional Bell Operating Companies (RBOCs) created by the MFJ from AT&T’s local phone divisions prohibited from most other lines of business offering most of the nation’s local telephone service. Long distance service was retained by AT&T, and was quite soon thereafter subject to competition, primarily from MCI and Sprint. The RBOCs and AT&T were regulated for the next 12 years by a combination of the court that enforced the MFJ under antitrust law, and the FCC, enforcing the 1934 Communications Act.

The Telecommunications Act of 1996 replaced this arrangement. Its goal was to open up the local telephony market to competition. This departed radically from the conception that dominated local telephony regulation throughout most of the 20th century, namely, that local telephone service was a natural monopoly to be regulated, but not to be subject to competition. Recognizing the interest of the incumbent local telephone companies to maintain their monopoly, the 1996 Act created a series of obligations on local exchange carriers in general, and on the RBOCs in particular, to cooperate with new entrants—their emerging competitors. In order to create a competitive marketplace the legislation offered the regional Bell operating companies (RBOCs) the opportunity to enter the long-distance market once they had demonstrated that the structural steps had been taken to irreversibly guarantee local competition.

Three mechanisms were provided to this end: mandating sale of telephone services at wholesale rates; leasing of access to unbundled elements of the network by competitors, and interconnection. Arbitration by the state public utility commissions, or by the FCC, is provided for by section 251 of the 1996 Act failing private negotiation between the parties. The basic idea underlying these three mechanisms is that it not be necessary for a competitor to have in place a full network, completely redundant to the incumbent’s network, before being able to offer competitive services. Any competitor can therefore enter by simply buying “minutes” and reselling them more efficiently—this is the resale component of the arrangement. It is intended largely to give an entrant an opportunity to build local reputation while preparing to build its competing facilities. Second, a competitor can buy, at cost, access to any network element of the incumbent. A competitor, for example, can simply deploy switches and some large trunks, but then lease lines from the switch to customers’ homes from the incumbent at cost. This is known as unbundling. Finally, a competitor that has a fully functional network to some customers in an area, has a right to interconnection—to be connected at cost to the incumbent’s network, so as to be able to provide its customers connections to all of the customers of the incumbent. The latter provision is crucial, because it allows an entrant to compete with an incumbent on an equal footing with regard to the value of telephone as a network good—a good whose value increases the more others consume the same good. If an incumbent were not required to interconnect, a competitor could only offer its customers the value of connecting to its other customers, and not to everyone on the planet. Denial of interconnection was a central strategy that the Bell System had used early in the 20th century to eliminate competitors in local markets, and again to prevent competition in long distance in the 1960s and 1970s. The 1996 Act sought to assure that it would not be used so again.

The incumbents fought implementation of the requirements to open their networks on a variety of fronts, largely through litigation. The most important of these efforts was the Iowa Utilities litigation. In the first iteration, the incumbents succeeded in winning much of what they wanted from the court of appeals, but were largely reversed by the Supreme Court in AT & T v Iowa Utilities Board 525 US 366 (1999). After the case was sent back to the court of appeals, that court upheld the FCC's general approach, but invalidated the specific requirements that the Commission had required of the incumbents. In May, 2002, however, in Verizon Communications, Inc., v. FCC, the Supreme Court finally reversed the court of appeals on all the items on which it had held for the incumbent exchange carriers, and adopted the FCC's cross petition, both in methodology and the reasonableness of the actual requirements. Six years after the 1996 Act was passed, with much litigation and stalling, the Supreme Court seems to have, for the moment, allowed the FCC quite significant leeway to regulate the rate and structure of the unbundling requirement.

The delay, however, was sufficient to allow political winds to change. The mood at the FCC seems to have turned from one vigorously dedicated to opening up the telephony platform to competition not only in voice telephony, but also in broadband/DSL delivery, to one that is much more hesitant to treat these types of services as common carriers. It is this issue--the way in which the measures, primarily unbundling--that the 1996 Act introduced to help competitors are applied to broadband over telephone infrastructure that matters most to understanding the problem of Internet access over this platform. Through a variety of approaches for high speed data transmission over copper wires, generally known as DSL, telephone infrastructure can be used to offer broadband Internet access. The question becomes whether the mechanisms adopted by the 1996 Act, as implemented by the Commission, can ensure that at least as far as broadband Internet access is deployed over telephone systems, access can be competitive and open. A glimpse at attempts that RBOCs have made to receive permission to offer long distance service suggests caution with regard to that question.

The 1996 Act had conditioned RBOC entry into long distance markets on FCC approval, in consultation with the US Department of Justice. Approval depended on a finding that the applying RBOC had indeed opened its local network to competition.

The first successful application, after a number had failed, was Bell Atlantic’s application to provide long distance service in the New York area, the most lucrative market in the United States. Bell Atlantic had taken a more cooperative tack than all its sister RBOCs, and had advanced further in opening up its network to interconnection and unbundling by competitors than had been the case in other regions. The 1996 Act requires authorization by the FCC in consultation with the Department of Justice. The FCC eventually approved Bell Atlantic’s application, a decision upheld by the Federal Court of Appeals for the D.C. Circuit. Nonetheless, the Department of Justice Evaluation of Bell Atlantic’s request, which counseled against granting Bell Atlantic's request, reveals the level of detail at which an incumbent can effectively thwart competition. The quality of the graphical user interface for receiving orders for entry, the speed of deployment of service personnel, and the design of order forms came under scrutiny, suggesting that for all practical purposes the opportunities for an incumbent to make the lives of competitors unbearable were infinite.

The lesson from these efforts—and the fragility of competition that requires access to a competitor’s infrastructure—need not be that competition or open infrastructure are impossible. But neither does it suggest reason to be sanguine that a formal requirement of openness will necessarily lead to competition and an open network. Even where access to broadband service operates on a common carriage basis, with explicit and extensive requirements for competitors to cooperate, there continues to be wide opportunity for incumbents to capture and control large segments of the Internet access market.

The "Appropriate Framework" NPRM generalizes the most important conceptual move in this area by tentatively categorizing broadband services over a wire as "information services" under the Communications Act. This categorization is immensely important, because there is no stautory category of regulated information services. The definition is in contradistinction to "telecommunications services" or common carriers, which are subjct to regulation. The explicit intent of the Commission in making this classificatory move was to allow itself to build an appropriate regulatory framework from the ground up, without being encumbered by trying to fit new services into old regulatory boxes. The result is likely to be to make any form of regulation very difficult. The definition of "information service" in the Act is:

Plainly, this is a very broad definition, and explicitly encompasses even online newspapers. The Commission will therefore have to be highly solicitous of the providers of broadband, and will likely regulate very lightly, if at all. The definition is also likely to be regarded by courts reviewing any regulation as indicative that the Commission believes these entities to be traditional media organs, which receive the highest First Amendment protection. The risk that a court will treat any form of access or unbundling requirement as functionally equivalent to a requirement that a newspaper devote some of its editorial page on a common carrier model is likely, an analogy that will surely prevent any access regulation.

An alternative approach would be to look at the functional architecture of the service, and to see that it is comprised of two distinct services--a platform service that in principle looks and behaves like a telecommunications service, and an ISP/value-added service that looks and feels like an ISP service. The latter is quite properly designated an information service. The former is not. Treating their combination as purely an information service is no more necessary than treating local and long distance service as a single "telephone service" category would have been. On this view, the Commission's definitions seem to run contrary to the original model of the 1996 Act, which sought to move away from regulation of types of companies towards regulation of specific functions that different companies combine. Had the Commission indeed defined broadband services as involving both a telecommunications service and an information service, it could easily have achieved its goal in this definition--to allow these services not to be encumbered by inappropriately applicable regulatory constraints developed for plain old telephone service. The Commission is empowerd, under the Act, to refrain from regulating an advanced telecommunications service when it deems that refraining would serve the public interest. Had it defined the platform component of the broadband service as a "telecommunications service," it could have started with the regulatory baseline applicable to common carriers and telecommunications services, and then decided to forbear from imposing regulatory components of that framework that would be inappropriate to broadband service. As things stand, however, the recent move to classifying all broadband over wireline as an information service tends to suggest there will be little regulation to force incumbents to open their networks to competitors.

The Incumbent LEC Broadband NPRM applies this framework to incumbent local exchange carriers, tentatively finding that these too are information services. The most important focus here is on the presence of substantial intermodal competition--that is, largely, that DSL has strong competition from cable broadband, which does not allow the incumbent telephony carriers to behave like monopolists.

The UNE Triennial review refers to a comprehensive review of all the Commission's approach to unbundling in the telephony infrastructure. The most important component for purposes of broadband is the question of whether the high frequency portion of a wire will be required to be unbundled. In an earlier order, the Line Sharing Order, the FCC had required the telephone companies to unbundle the high frequency portion of their wires. This is a metaphorical way of referring to the following technical situation. Voice telephony encodes the voice it carries as a series of low frequency electromagentic signals. Data communications encode data using high frequency signals. We are used to thinking of "frequencies" as a physical thing in the airwaves, but that is entirely false. Frequency measures a property of any electromagnetic signal, irrespective of whether it travels in the air or in a wire. Because voice and data use electromagnetic signals with such different characteristics, they can use the same wire without confusing equipment connected to the wire as to which signals are voice, and which are data. Competitors of the incumbents wanted to rely on this fact to purchase from the incumbent LECs the right to send data over the same wires that the incumbents were sending voice. This would enable them to sell data services without buying a whole dedicated line. The incumbents wanted to sell the competitive entrants a much more expensive "network element"--the whole wire. The Commission conceptualized this as though there was a pipeline with different frequencies in it, and required incumbents to unbundle the higher frequency portion. From a policy perspective, this was clearly the right choice if the correct goal is to lower the entry costs to competitors of the incumbents, without harming the incumbents (beyond the harm they suffer from the cheaper introduction of competition). In the UNE Triennial Review, the FCC is requesting comment on whether it should reopen this requirement. This inquiry may, however, have been rendered moot by a decision rendered in late May 2002 by the D.C. Circuit, which invalidated the Line Sharing Order as unreasonable. It is not entirely clear that this opinion quite fully squares with the relatively broad discretion and deference that could be read into the U.S. Supreme Court decision in Verizon, although it is not directly in conflict with that holding.

The conclusion to all this is, unsuprisingly, that things are in flux in the telephony-based infrastructure for broadband Internet access. While the Surpeme Court seems to have recently given the Commission broader powers to use unbundling as a means of facilitiating entry by competitors, both the Commission itself and the D.C. Circuit are actually moving away from regulating the use of shared facilities to promote competition among competitors using the incumbent telephony infrastructure. Instead, they seem to be relying more heavily than ever on the presence and long term beneficial effect of intermodal competition from cable.

At least 80% of US households have access to cable service. This infrastructure is currently being upgraded from the coaxial cable - designed for the one-way transmission of video programming - to hybrid fiber co-axial for the purpose of broadband Internet access. Data carriage capacity is higher for cable broadband than comparably priced DSL services. It is also available more widely throughout the cable service area, because DSL availability is limited by distance from a central office. Cable transfer speeds, however, fluctuate in accordance with the number of users at a given time, while DSL is a dedicated line and more predictable. Cable Commission’s report Broadband Today, though slightly dated, still provides a good starting point to understand the forces operating in this area.

The open access debate arises from the cable network owners' practice of bundling cable modem access and Internet service provider (ISP) selection. They had initially signed exclusive contracts with affiliated ISPs, Excite@home and Roadrunner. More recently, these cable providers have been providing their own systems. Cable providers, unlike the telephone companies, are regulated under Title VI of the Communications Act. Most importantly, this means that they have never been not subject to common carrier restrictions, and the various requirements imposed on the incumbent local exchange carriers to cooperate with competitors do not apply to them. Independent ISPs feared that exclusion from the cable network would leave them incapable of competing in quality of service with the cable operators’ affiliated ISPs. The DSL providers, in particular the incumbent local telephone carriers, objected to the fact that they must provide access to their network to competitors, whereas the cable companies need not. In opposition to the demands of open access advocates, cable network owners and their allies argued that imposition of forced access upon them will result in slowing the availability of high-speed access. The cable infrastructure requires adaptation to be used interactively and their claim is that in order to finance the adapatation they need exclusive control. They have also asserted that technical and security questions impede the opening of access to other ISPs. The politics of the debate are complex, and heavily influenced by interest. AOL, for example, was initially a major part of in the OpenNet coalition’s efforts to persuade the FCC to impose open access requirements on cable service providers. It reversed its position after it announced its merger with Time Warner. The merger, however, provided the first opportunity to impose limited open access requirement as part of the merger approval.

A central aspect of concern with cable access is whether an owner of physical infrastructure, that also controls Internet access, can to a large extent affect the content of the information its customers will see, and how they will use it. (These concerns were given grounding and heightened by a White Paper from Cisco extolling the benefits of such a strategy as part of a sales pitch offering the tools for its execution “Controlling Your Network - A Must for Cable Operators”). Over and above the concerns about parity in competition, there is a concern over the eventual bifurcation of the Net, with high-end access available to commercial content, and low-level access for non-commercial content. The emergence of commercial services such as Akamai accelerating access to e-commerce and commercial sites points to the tendency towards bifurcation present even in the absence of infrastructure-based handicapping. With an added layer of preference for commercial content by the infrastructure itself, we could find ourselves with a medium much more like the mass media of the 20th century than the Internet of the 1990s. Elements of the approach are evident in the self-description of At Home's network. Restrictive end-user contracts from cable companies indicate that they are concerned to avoid competition not only from their industry competitors, but also from production of content by users. End-user contracts forbid media streaming of more than ten minutes duration and the operation of a server over the cable modem network Other means by which proprietors can exploit their position include selective caching of content produced by them or their partners - so that it is easier to access and cleaner to watch - and the collection of data on customer use patterns.

The general response at the national level to efforts to open access to the cable infrastructure was cool. The FCC, for example, refused to impose open access requirements upon AT&T in reviewing its mergers with TCI and MediaOne. (See Lessig on access in the MediaOne Merger.)

The first successes that open access advocates enjoyed were at a local level. Cable companies are regulated as much, if not more, by local franchising authorities as they are by the FCC. Portland, Oregon was among the first locales whose local franchising authority mandated open access by ordinance. These requirements have been overturned in court in holdings of two varieties. In the case of Portland an interpretive approach to the 1996 Act deemed cable broadband access not within the definition of cable service. In Broward County, Florida, a Federal District Court took a constitutional approach, and found that broadband access was a cable service, but that imposition of conditions on its provision violated the First Amendment.

47 USC Section 541(c) of the Communications Act states: "Any cable system shall not be subject to regulation as a common carrier or utility by reason of providing any cable service." But in many instances, the franchisee needs the franchising authority's approval in case of sale. This, at least in principle, opens the possibility of the imposition of conditions upon the grant of that approval under 47 USC 537, and under 47 USC 533(d)(2) which states: “Nothing in this section shall be construed to prevent any State or franchising authority from prohibiting the ownership or control of a cable system in a jurisdiction by any person … (2) in circumstances in which the State or franchising authority determines that the acquisition of such a cable system may eliminate or reduce competition in the delivery of cable service in such jurisdiction.”

When AT&T bought TCI, it sought to transfer to it the cable franchises that TCI had previously held. In Portland, this was used as an opportunity to get AT&T to comply with the city’s requirement that AT&T open up the cable broadband infrastructure to competitors to offer Internet service. AT&T rejected Portland’s open access requirement, and its request to transfer the franchise was accordingly denied. AT&T sued the city, seeking invalidation of the ordinance. The District Court upheld its legality but was overruled by the 9th Circuit on appeal. The Court of Appeals rejected the classification of cable broadband access as a 'cable service', distinguishing it from the 'one-way transmission' of programming with highly limited subscriber interaction’. Consequently Portland did not have any legal basis to regulate the delivery of the service under the franchise agreement. Indeed the court held that no cable franchise was required for delivery of broadband services. The court decided that broadband internet access was a hybrid service, constituted in one part - the 'pipeline' - by a telecommunications service and the other -as an ISP - by information service. Accordingly, regulation of the service by the cable franchising authority was struck down.

Broward County in Florida adopted an open access ordinance requiring the cable franchise to provide access to its broadband network 'on rates, terms and conditions at least as favorable as those on which it provides such access to itself.' The ordinance was the subject of a successful first amendment challenge for interference in cable broadcaster's power to determine programming available on their systems. Judge Middlebooks likened open access to the right to reply statute found unconstitutional in Miami Herald Pub. Co. v Tornillo, and dismissed the argument that cable companies enjoyed a bottleneck monopoly over access to the Internet on the grounds that most users connect over phone lines.

One of the determinative factors in both these cases was the FCC’s unwillingness to support open access. An important, though in some sense little stated, concern faced by the courts was whether such a question of national communications policy ought to be decided on a local basis by hundreds of franchising authorities. The sustained public pressure seemed to yield some change in national policy when it came time for the relevant federal agencies to approve the AOL-Time Warner merger. In a sense, that merger crystallized the concerns with a fully integrated cable broadband infrastructure. The nation’s largest ISP bought one of the two largest cable operators, and one of the nation’s largest content libraries. Together, the picture was one of a fully integrated network. Consumers would be presented with an infrastructure and an interface controlled by one corporation. This corporation also controlled many media properties and would have incentive to use the opportunities control over the infrastructure gave it to focus users’ attention and time on its content to the detriment of others. To obtain FTC approval of its proposed merger with AOL, Time Warner entered into a five year consent agreement to allow access by rival ISP Earthlink to their broadband network in 70% of their markets, prior to making available AOL’s broadband ISP service (November 2000). In addition it must contract with at least two other ISPs within ninety days of the launch of AOL’s service. In the remaining markets the company must contract with three non-affiliated ISPs within ninety days of the launch of AOL’s service on terms and conditions as favorable as those provided to AOL and affiliated ISPs. Network flow data and interconnection points must be provided to affiliated and non-affiliated providers on a non-discriminatory basis. While this is only a limited requirement, and only applies to AOL Time Warner, it was interpreted at the time as an important first instance of such a requirement. Following this decision, the FCC issued a Notice of Inquiry process regarding access to high-speed Internet access over cable, focused in large measure on the question of whether something like the moderated open access requirement imposed on AOL Time Warner should be generalized.

The most recent FCC activity in this area seems, however, to begin a tentative move away from open access. As part of the series of rulemakings issued in early 2002, the FCC issued a Declaratory Ruling and NPRM that makes two important moves away from open access in cable. Both involve the declaration that broadband internet access over cable is not a service that combines telecommunications services with information services, but is, rather, purely an information service. The first, direct, effect that this declaration does is securely to federalize the issue of access to the cable broadband platform. By defining cable broadband as an interstate information service, the FCC asserted sole jurisdiction over it, and thereby preempted action by local franchising authorities. Given that these authorities were the primary sources of efforts to regulate cable companies broadband provision, this decision was an important conclusion to the move to federalize this question already implicit in AT&T v. Portland. The second, probably more important but less direct effect that this declaration had was to make it very difficult for the Commission to adopt a seriously invasive access model. As an information service, cable broadband is categorized in a very broad category that could apply to many things, in particular online newspapers, that could never consitutionally be required to open their facilities to others. The result is that AOL Time Warner is subject to some access regulation because of the merger. If other cable providers similarly merge, the Federal Trade Commission might impose some access requirement, as it did in the AOL Time Warner merger. But the FCC at this stage seems to have decided that competition from the Incumbent LECs will suffice to secure non-monopolistic behavior.

The result of the cable decisions, given the parallel move away from requiring extensive unbundling and forced sharing of inputs on the telephony side, is that the FCC is preparing to make peace with a duopoly situation based on pipeline ownership. Almost all current broadband over telephone lines is offered by incumbent LECs. Almost all cable broadband access is offered by ISP affiliates of the cable operators. Each pipeline owner will, for all practical purposes, enjoy a monopoly over its own pipeline. Even assuming the best of all possible worlds, where the reach of each pipe is equal and equally comprehensive, we are left with a duopoly. Two are better than one, perhaps, but to describe such a system as achieving the open, competitive system that the Telecommunications Act of 1996 envisioned would be something of an exaggeration.

While many of the arguments about open access are concerned with traditional antitrust concerns—consumer welfare, quality and price of service—and with capacity of users to speak, Mark Lemley and Larry Lessig formulated a different argument in support of access. They focus on the potential for cable access bundling to wipe out the layer of ISPs that have provided much of the innovative impetus for the flourishing of the web. In addition, they identify this as a step towards undermining the end-to-end design principle that undergirded the network architecture of the Internet, which ensured maximum accessibility by placing all intelligence at the end of the network rather than at its core, where, it has been said, stupidity is its strength.

Many of the problems that plague the deployment of high-speed internet access over the legacy wired infrastructure have to do with the very high costs of construction and the difficulty of acquiring the necessary rights of way to deploy the system. One infrastructure for communications that does not suffer from similarly high set-up costs, at least not necessarily, is offered by wireless communications. At the moment, the limited, controlled nature of wired infrastructure is being replicated in the air by licensing policies that limit the number of providers who can offer Internet access over the air. But this regulatory creation of a controlled-infrastructure model for wireless communications is technically obsolete. (for a detailed analysis go here.)

Until the last decade, the universe of options for regulating spectrum was technologically limited. Given the relative crudeness of reception devices that could be made cheaply enough for consumer markets, the generally accepted understanding was that the only way for a signal from a transmitter to be received by a receiver was for the transmitter to be louder than all other sources of radiation in a given frequency. This meant that if two or more transmitters tried to be "heard" over that frequency, there would be "interference"—neither would be sufficiently louder than the other. This basic technological-economic fact limited the menu of institutional options open for government in regulating wireless transmission. Someone had to be given the exclusive right to transmit loudly over a given narrow frequency, at a given time, in a given location and power. Policy debates in the area of spectrum management have therefore focused for almost half a century on whether that someone should be chosen by licensing or by auctioning, and to what extent the use to which they put their license should be determined by the FCC as opposed to by the licensees responding to market signals.

Advances in computer processing, network communications, and wireless transmission technologies, mostly spread spectrum and software defined radios, have now made possible a third alternative. Receivers and transmitters can be made intelligent enough to share spectrum. Multiple users can and do share wide swaths of spectrum simultaneously. Allocation and assignment are achieved on a packet-by-packet basis, using equipment-embedded protocols rather than organizational decisions by a licensee or spectrum owner. "Spectrum" is "managed" not by licensing, but on a license-free model, where all equipment that complies with some standard for sharing--at a minimum complying with a power limit--is permitted to operate without a license.

The technological shift has begun to bring about some initial changes in spectrum policy, and somewhat more significant developments in the equipment and service markets. Traditional spectrum policy permitted operation of low power devices—such as cordless phones, garage openers, microwave ovens—in very narrow bands of spectrum, like the now much-used 900MHz band and 2.4GHz band. These were generally dumping grounds for sundry useful low power emitters, rather a focus of communications policy. As the 1990s progressed, the FCC permitted operation in the U-PCS band and the U-NII band (see the FCC's Unlicensed National Information Infrastructure Order of 1997), as two instances where the FCC self-consciously permitted operation of low-power devices without a license, with the expectation that communications services would be offered in reliance on a license-free model. The problem with these was that license-free service was permitted and designed not as a primary focus, but primarily by reference to its potential interference with licensed incumbent services. With the U-NII Band in particular, the 300 MHz in which license free operation was permitted, was sliced and regulated seemingly guided largely by a concern for incumbent licensed services, not for what would optimize operation of the license-free devices themselves. More recently, the FCC has begun to look more seriously at flexible transceivers using license free spectrum as a serious direction. Most important in this regard are the Commission's adoption of the Ultra-Wide Band order and the permission to manufacture Software Defined Radios. These are substantial moves towards permission for equipment manufacturers to go ahead and experiment with wireless communications that utilize a combination of techniques to break away from the old model of wireless communication and create an entirely new model.

Market actors have been less reticent about license-free spectrum, and some have taken advantage of the limited spectrum that is available to develop quite significant applications, utilizing a variety of standards and approaches. Apple’s Airport is a particularly visible example, but many PC producers are now integrating license-free wireless networking into their computers, and some equipment manufacturers—like Nokia—have developed equipment that could be the basis for large-scale high speed data networks owned by no one. Actual networks built around license-free equipment have emerged on a nonprofit basis around the world.

The year 2001 seemed to have been for license-free wireless what 1999 was for free software. Equipment using standards developed for the scraps of spectrum in which license-free operation is permitted are beginning to become ubiquitous. Different standards--802.11, Blue Tooth and Home RF, for example, sought acceptance in devices capable of transmitting at much higher speeds than those currently available over cable or DSL, using spectrum owned by no-one. In conjunction with TCP/IP protocols, these devices can provide the last 10 miles, or mile, or at least 100 feet connection to the Internet, and make it mobile to boot. If this should happen, it would radically alter the lay of the land on Internet access. By late 2001 and erly 2002 this possibility began to be appreciated even in some media accounts.

To get a flavor of the possible, you can browse around work done by Dave Hughes, a researcher who conducted numerous field tests connecting schools in rural areas, you can read about them here, or listen to him give a presentation. Another important locus of implemenation projects is the Dandin Group, headed by Dewayne Hendricks. Important and accessible explanations are availble in Reed's Locus. The intrepid may wish to examine the technical detail provided by the FCC Technological Advisory Committee's working group on spectrum management (SWG). Hardware manufacturers such as Nokia and Apple are driving the evolution of these forms with systems such as Wireless Rooftop Routing (technical papers) and Airport. You can also read the CarNet proposal for a mobile wireless network. You can read about policy proposals generated by these technical possibilities. Eli Noam has proposed a market model which would clear competing uses based on willingness to pay a price which would vary depending upon the volume of traffic on the network at the time. Benkler argues for a system more akin to the Internet protocol itself, without reference to payment at all, as the benefits of such an arrangement to democracy and autonomy outweigh its functional deficiencies-the wireless spectrum can provide a commons ‘network of last resort’ coexisting with the traditional alternatives. Critics, such as Thomas Hazlett, are skeptical of the capacity of wireless to provide broadband requirements. Given that broadcast regulation by the FCC has always been predicated upon the assumption of scarcity and a consequent need to limit users so as to prevent interference, the implications of spread spectrum technology are profound. Some, like Noam, and Benkler and Lessig, have argued that refusal to permit widespread use of license-free devices could make FCC regulation of radio unconstitutional.

“In digital networks, this is the key layer: this is where network configuration is defined, where interconnection between separate physical networks is made possible or prevented, and where co-existence of various service providers is permitted or denied.” Bar & Sandvig, "Rules from Truth: Post-Convergence Policy for Access"

While obstacles to market entry at the physical layer derive from the high costs and time required to establish alternative facilities, soft infrastructure erects barriers through a series of positive feedback that permit one company to lock in an early advantage and make entry by competitors difficult. As the amicus economists’ brief in the case of Lotus v. Borland describes, the nature of software markets means that products whose initial success may be serendipitous and contingent can capture a market and exclude even products that are intrinsically better by some functional measure.

Moreover, markets in software that performs discrete functions are often closely related to each other. Very often providing a given functionality can depend on access to another functionality performed by software in a market already captured by someone. This characteristic provides opportunity for a company that has become dominant in one market, to leverage that position to extend its monopoly into markets that rely on access to the segment of the logical layer that it controls. This was the argument the government made in the case against Microsoft and its Windows OS, vis-à-vis web browsers. The District Court’s findings of fact in the Microsoft case document in detail the strategies used to produce this effect: bundling of the web browser with the OS, control over the active desktop, and manipulating user reliance on default settings. The Microsoft case illustrates that in principle, antitrust policy can work in 'tippy' markets by enforcing interoperability, which largely parallels the requirement in telecommunications markets of enforcing interconnection. Copyright law could attenuate this danger to some extent, though not completely, by privileging reverse engineering and refusing protection to interface aspects essential for the viability of new market entrants–ensuring backwards compatibility and horizontal interoperability, as the court in Lotus v Borland did.

The insight is by no means limited, at least in principle, to Microsoft and operating systems. Prior to the approval of their merger with Time Warner, America Online resisted opening its Instant Messaging systems to interoperability with other instant messaging systems. AOL IM and ICQ dominate the IM market with 90% of installed clients. AOL's refusal to interoperate with other providers means that users must run parallel instant messaging clients to communicate with ‘buddies’ using services offered by other providers such as Odigo, iCast, Yahoo and Microsoft. AOL actively obstructed competitors' attempts to interact. It argued that the swift growth of their competitors' user base demonstrates the absences of any anti-competitive threat, and that control over interoperation was necessary to assure integrity of their service. The success of IM has attracted numerous application developers aiming to capitalize on its potential as a content delivery platform, incorporating telephony, video conferencing and a variety of point-of-presence driven services. Control over the IM platform would, one might be concerned, give AOL similar opportunity to exert power over these applications that Microsoft’s power gave it vis-à-vis application developers who needed access to its operating system. Subsequent to the FTC’s imposition of approval conditions on the AOL/Time Warner merger, the FCC appended a requirement that AOL provide access to IM should the service be synthesized with cable broadband to offer ‘advanced instant messaging'. So long, however, as IM remains limited to phone and DSL lines interoperability will not be mandated and AOL will retain full control. An example of the effect on this on the compatibility of new programs can be seen in the fact that Jabber, the open-source project that provides an instant messaging functionality, is compatible only with Yahoo and MSN's instant messaging, not AOL's AIM and ICQ. One should, however, remain somewhat cautious about the claim that AOL's instant messaging programs will in fact hold bottleneck control. Firms attempting to offer client-side interoperability suggest that, with some initial start-up costs, it is possible that AOL's hold will be substantially less complete than Microsoft's. This is an example.

One solution to the problem of closed platforms at the logical layer is to develop open public standards for how applications operate and interoperate with each other. A standard is a codified way of performing a function. It allows anyone who needs to perform the function as part of a function being provided by their software to do so based on the standard, and be confident that the application will work with equipment and other applications that similarly adhere to the standard. If many who need to use the functionality can participate in formulating the standard, and it is open for anyone to read and understand fully, then different applications, created by different companies or individuals can all interoperate without giving anyone the power to control communications that depend on the functionality. Organizations like the IETF, W3C, The Internet Society and the IEEE provide for standard setting that, when followed, provide an open framework for software, whether or not proprietary, to interoperate. A more radical remedy, more complete in its promise to maintain an open logical layer, is the placement of free software or open source software in critical junctures of the logical layer. Already, the Apache server software that most web servers use, and Perl, the programming language used for most scripts and many other cross platform functionalities on the Web, are free software. In September 2000, the President's Information Technology Advisory Committee produced a report on open source recommending concerted federal support. As more of the core of the logical layer is free software, open for all to see, use, and modify, the logical layer becomes an open layer that cannot be manipulated by any one entity to control Internet access.

While most of the focus in this module has been on the structure of access to the Internet and the ways in which it does or does not give infrastructure owners the power to control use of the Internet, an important social policy concern with Internet access has been the distribution of access across class, race, gender, age, and education attainment. Treated mostly in the latter part of the 1990s as the problem of the Digital Divide, the problem is one that in traditional telecommunications policy has been treated under the umbrella of “universal service.”

At inception, universal service referred to the need for a unified service in a context of a fragmented phone system without interconnection requirements, rather than a conscious policy to promote residential telephone penetration and rural telecommunications facilities through subsidies gleaned from long-distance. It was introduced as a concept by AT&T’s President, Theodore Vale, in 1907, largely as a justification for the Bell System’s push to monopolize telephone service throughout the United States. According to Milton Mueller, as competition began to enter the long-distance market in the 1970s, Bell invoked universal service as part of its mission for reasons of political expediency, namely to retain their position as a regulated monopoly.

Coming from these inauspicious roots, the solution developed to the problem in the Telecommunications Act of 1996 was incomplete. The FCC operates targeted schemes to bring access to specified groups under programs like Lifeline and Link-Up, which provide plain old telephone service of some degree to the poorest users. E-Rate, is the program aiming to bring schools on-line, and the one most closely related to Internet access. Under 1996 Act the FCC and a joint federal-state board is supposed to develop a standard of "an evolving level of telecommunications services" that is to be reconsidered and updated periodically, that would be subsidized for poor users. This evolving definition for universal service is intended to comprise services subscribed to by a majority of American, but inclusion of Internet access for the general user remains undetermined. Any service brought within the definition becomes eligible for subsidy. The FCC released a first definition in May 1997 and presented a detailed Report to Congress on the functioning of universal service in 1997. Nonetheless, while much political airtime has been spent on the Digital Divide, relatively little by way of active subsidization or implementation of ubiquitous access has in fact been achieved beyond the declining price of computers and Internet connections.

1. AT&T Corp. v. City of Portland. Imagine that you are a Supreme-Court justice hearing arguments from AT&T and the City of Portland on appeal. You know that the Federal Trade Commission has required AOL Time Warner to open its network to some competitors, and that the FCC is considering the issue more broadly. How would you rule and why? Do you think these issues should be decided at a national or local level? How do you see your decision affecting the architecture of Internet access over cable?

2. You have just read papers describing Nokia's Rooftop Networks. Imagine that you are: (a) a staff member at the FCC's Office of Engineering Technology responsible for spectrum management decisions; or (b) a lawyer for an industry group of wirless providers who have spent billions of dollars purchasing spectrum licenses at auctions in the past few years. Imagine that the FCC is considering whether the UHF bands, that will be released by television stations in 2006, should be auctioned off to providers of 3G wireless services or retained for a license-free spectrum. What would your arguments be about whether the spectrum should, or should not be auctioned?

3. Local exchange carriers often argued for regulatory parity. Why, they asked, is it fair to require them to provide their competitors in high-speed Internet access equal access to their facilities, without similarly requiring cable companies to do the same? This fairness objection could be resolved in one of two opposite ways. Cable companies could be required to offer competitors unbundled access to their broadband platform, or telephone companies could be relieved of their obligations. An argument in favor of relieving the telephone companies of their obligations is that having two wires in each home--the DSL provider and the cable provider--would suffice to assure that consumers benefit from competition, while leaving both telephone companies and cable operators enough incentives to make the huge investments necessary to bring broadband to the home. The presence of competitors who can free ride on the investments of the infrastructure owner at cost dampens incentives for investment in infrastructure. If, this argument goes, consumers value highly being able to access many internet service providers, then either the telephone company or the cable company will likely open their system up to other ISPs as a competitive strategy against the other infrastructure owner, to capture those consumers who value choice of ISPs. The contrary argument is that neither cable nor telephone companies should be allowed to exclude competitors, and that there should be competition not only between two wires, but also within each, by regulating them so as to make space for any number of ISPs to compete with each other and with the infrastructure owner. Only thus, goes this argument, will we get enough competitors. These are precisely the questions that the FCC must face now, as it decides how to regulate the newly-defined cateogry of "information services" into which it has now put both cable broadband and DSL. What do you think?

4. The District Court in the Broward County case held that requiring a cable operator to open its network to competing providers violated AT&T's first amendment rights by forcing them to carry the speech of another. Public interest advocates have argued that, to the contrary, open access is mandated by the first amendment, because it is necessary to enable individuals and many diverse groups to speak, and the permitting a cable company to control Internet access over its infrastructure is what harms speech. Which is the more persuasive argument?

5. One approach to building high-speed Internet access is to construct municipal networks. Chicago, for example, is working on what it calls CivicNet. Some municipalities are building on their existing network of sewers, working with a company called CityNet, which uses robots to pull fiber conduits through sewers to every home. Deployment has already begun in a number of U.S. cities, and the company is seeking contract to do the same elsewhere, beginning with a contract to pull fiber through the sewers of Vienna. Is this a solution to the Internet access problem? Is it appropriate for government to "compete" with private broadband infrastructure providers? Will it help or harm speech and innovation to have local governments owning broadband municipal networks?

6. What would you say to the argument that the debate over Internet access and the precise architecture of the network and the market are really only the concern of a small privileged class, both nationally and internationally. The really important issue is distribution of access. With so many people on the planet lacking the most basic telecommunications facilities, it is much more important to make sure that more people get some Internet access than to spend all this political and regulatory energy on the precise details of how Internet access is configured.

7. Should universal service impose an obligation on federal and local government to finance public provision of fiber optics and wireless spread spectrum devices for all? Dave Hughes argues that at least the E-Rate program should be usable as a public subsidy for purchasing wireless equipment for building an unowned infrastructure. What do you think?

8. In approving the AOL Time Warner merger the FCC deferred open access requirements to future services combining Instant Messaging and cable broadband to deliver ‘advanced instant messaging’. Should the FCC have paid more attention to interoperability in messaging systems today, rather than focusing on the future? Is there a concern that customers will more or less become committed to AOL’s software, so that opportunities for competition later will become moot because the market shall have tipped? Would such a development affect, perhaps, the viability of services built around instant messaging facilities, like Madster?

9. Lemley and Lessig contend, that: “If a regulated entity threatens to force the adoption of an architecture which is inconsistent with the Internet’s basic design, and if that action affects a significant portion of a relevant Internet market, then the burden should be on the party taking that action to justify this deviation from the Internet’s default design. The presumption should be against deviating from these principles….As with any principle, these presumptions should apply unless there is clear evidence that displacing them in a particular case would be benign.” Is it the proper role of regulation to adopt and support a particular architecture, even if it is open and conducive to innovation? Can policy afford to be otherwise, if Lemley and Lessig are correct in their predictions?

Bar, François, and Christian Sandvig, "Rules from Truth: Post-Convergence Policy for Access", paper presented at the 28th Annual Telecommunications Policy Research Conference, Arlington VA, Sept 23-25, 2000.

Jerome H. Saltzer, "Open Access" is Just the Tip of the Iceberg, October 22, 1999 FCC Cable Service Bureau Homepage for America Online, Inc. and Time Warner, Inc. Proposed Transfer of Control.

Yochai Benkler, A Speaker’s Corner under the Sun in The Commodification of Information: Political, Social, and Cultural Ramifications (N. Elkin-Koren, N. Netanel, eds.). Kluwer, 2000.,