Copyright © 1995
Wake Forest Law Review Association, Inc.; Henry H. Perritt,
Spring 1995
ACCESS TO THE NATIONAL INFORMATION INFRASTRUCTURE
30_WAKE_FOREST_L._REV._51.htm
Henry H. Perritt, Jr. [FNa]
How the law addresses the issue of access rights to the National Information Infrastructure is bound to be critical in the radical landscape of cyberspace, where users play interchangeable roles and every consumer may also be a producer. Professor Perritt provides a detailed overview of existing legal doctrines which may provide guidance in structuring access rights, including common carrier, contract, antitrust, and constitutional law precepts. He concludes that given the novel nature of the Information Superhighway, such traditional sources of law provide only limited guidance. Legally mandated access duties may be required in certain areas, but that mandate should be narrowly tailored and restricted to interoperability and compatibility concerns. Externally imposed legal access duties may be unnecessary where competitive market forces act as a natural guarantor of access rights.
*52 INTRODUCTION
The National Information Infrastructure (NII)-consisting today of distinct components including the telephone system, broadcast and cable television networks, libraries, bookstores, remotely accessible databases, and the Internet-will soon converge. New broadband-switched networks [FN1] with digital connections will bring the NII into homes and public facilities. Increasingly, this revolutionary medium will have the potential to provide an electronic market for information, and an electronic town hall.
For this vision to be realized, however, a truly uniform infrastructure must exist, not merely a multiplicity of electronic boutiques, none of which is connected to the others. It is not yet clear how legal standards and governmental policies best can be structured to encourage the development of such a uniform infrastructure. One possibility, abstractly speaking, would be to impose legal obligations on parts of the whole to provide access to other parts of the whole. Conversely, it is conceivable that market forces will provide the best stimuli to the development of a uniform NII, considering that new technologies and consumer preferences are difficult to predict and therefore regulation may fail. Determining the appropriate boundary between legal and market forces is the central challenge of NII policy. This article addresses how that boundary should be defined with respect to the specific issue of access rights.
Law emerges from human cultures, and the NII represents a convergence of five different cultures: Telephone systems, broadcast media, textual*53 media, personal computing, and the Internet. Broadcasters, entertainment producers, and cable carriers are most familiar with high-bandwidth, largely analog, and mostly one-way networks. [FN2] Telephone companies, working within the confines of a detailed regulatory scheme, have recently begun to escape from an environment dominated by analog signals, [FN3] and have achieved nearly perfect interoperability and universal service at affordable prices. Text publishers, librarians, and government information enterprises know a context in which networks are multipoint and mostly digital. Personal computer (PC) hardware and software producers and users have known a world in which distributed computing, free markets, and autonomous individuals or small groups are the norm, as free-market entrepreneurs have wrought a revolution in computing and have brought high levels of technology to millions of small businesses and ordinary people, with little guidance from the government. Internet users are accustomed to being able to engage in information transfer and significant interoperability through a wide-area network linking dissimilar computing systems; [FN4] for them, governmental guidance was a great success, taking inter-networking standards from the lab to the marketplace.
Within these five cultures, three conceptual models for information transfer have maintained distinct characteristics in the past but now are merging: broadcast, messaging, and database models. The broadcast model historically was one-way, analog, and entertainment-dominated. But currently, the broadcast model is being deflected toward the messaging model by digital video and more demanding switching requirements for larger numbers of channels.
The messaging model historically was dominated by text. Today, however, advances in desktop computers have facilitated the integration of graphical images and digital audio clips into textual objects and the easy communication of such integrated objects through messaging networks. Such integration puts pressure on bandwidth because graphical objects and digitized audio require much more bandwidth than text. Integration also puts pressure on standard setting because there are many different ways to represent graphical images and audio signals, and to integrate them with text.
Like the messaging model, the database model historically was dominated by text. In the past it was the least open, or the most proprietary, of the three models. But because no single database provider has been able to control all of the information that its customers might want, recently *54 there has been pressure to open up these architectures, primarily through information gateways, to other holders of information. Client-server architectures and disaggregation of information value [FN5] may produce substantial efficiency gains in this model.
The Internet, albeit in a primitive way, is a hybrid which illustrates combinations of models. Usenet [FN6] newsgroups combine the database and messaging models. Listservs [FN7] combine the messaging and broadcast models. World Wide Web, [FN8] gophers, [FN9] and anonymous FTP servers [FN10] combine the database and broadcast models in the sense that one can "broadcast" or publish by posting an information object in a place known to be frequently and widely consulted.
Ideally, no single one of these models or cultures should dominate the NII. Each of the three models and five cultures has valuable attributes to offer. The interoperability and universal service features of the telephone system are worth perpetuating. So too are the connectivity, interoperability, and open architecture features of the Internet. The enormous market which has been tapped by broadcast media may stimulate innovative technology investment and lower prices. The worthwhile traditions of artistic creativity, political discourse, and packaging of information which have historically been associated with textual media should be continued. The PC revolution has demonstrated the superiority of the free market and entrepreneurial effort in facilitating the development of good, affordable software for ordinary people. The challenge for policy *55 and law is to retain the best aspects of the different cultures as they converge into one.
This article draws the conclusion that traditional legal doctrines provide only limited guidance in defining the contours of access rights for the NII. Statutory common carriage is too detailed and rigid a doctrine. There is, however, some potential for common law concepts to aid in the development of a unified law of NII access rights. Common carriage doctrine could be developed to address certain potential access disputes. Collective negotiation of standard service terms could be governed effectively by a contract regime. Further, antitrust essential facilities doctrine may provide a useful way to ensure access to governmental information. Beyond such common law controls, legal intervention in the form of limited statutory action may be required. Statutory action should be unnecessary to ensure access to most parts of the NII because competitive market forces are likely to prove sufficient. With respect to certain aspects of the NII, however, legally mandated access is appropriate. Such a mandate should be narrowly tailored and limited to interoperability and compatibility concerns. [FN11]
I. FOCUS: ACCESS BY SUPPLIERS OR BY ULTIMATE CONSUMERS?
Access rights and universal service issues are more difficult to distinguish in the NII context than in the context of prior commercial technologies. The NII in many ways resembles the Internet, in which every consumer of information is also a potential supplier. [FN12] Universal service, which has traditionally been considered in the context of telephone systems, now has implications for producer access to channels of distribution because the end users are also potential producers. Historically, producer *56 access to channels of distribution was primarily considered in the context of broadcast and cable television, which was predominantly a one-way medium. In the distributed, Internet-like NII architecture of the future, even large producers almost certainly will use the same infrastructure both to acquire their raw material and to deliver their product. Thus, they will be both consumers and producers through the same network connections. One may be skeptical that significant markets will exist for the types of casual content produced by small consumer/producers. In that specific context, the quality and finding-and-retrieval problems may be too great. But even if real markets distinguish to some degree between end user/consumers and somewhat larger producers, it will be difficult for the law to distinguish, as a threshold matter, between producers and consumers, since the two categories are blended together in the NII context. [FN13]
The convergence of consumer and producer roles has significant implications for the nature of legal regulation. Access to the NII in order to put something in becomes as important as access in order to take something out. As a general proposition, compelling a provider to allow someone to put something in potentially involves much greater impact on product design and facility control than compelling the provider to let someone take something out. For example, forcing a supermarket to allow anyone to shop there represents a certain amount of interference. But that is far less interference than forcing the supermarket to provide shelf space for a particular product. Similarly, the law interferes far less with the entrepreneurial prerogatives of an information services provider by forcing it to allow someone to extract information from a preexisting system than by forcing the same provider to accommodate a new type of information content as something supplied or put in. The second type of access may force system redesign, while the first type is unlikely to force such changes. At most, ensuring the first type of access, information extraction, would merely require an increase in capacity.
The way out of this dilemma may lie in specifying supplier-oriented types of access controls that are less intrusive than the alternative of requiring providers to accept many types of content. Requiring one supplier to be compatible with another-that is, requiring interoperability-does not impose as much on the obligated entity, as requiring providers to accommodate incompatible suppliers. Further, if compatibility is not designed into a supplier's facility from the outset, translators always can be written and used at the cost of some type of performance penalty.
II. SURVEYING THE INTERESTS INVOLVED
Access rights involve the interests of three types of network service users, in addition to the interests of the providers who may owe access duties. End users may assert a right to have initial connections made, as *57 well as a right not to have existing service terminated. Suppliers of content or value-added services may assert rights to have their content or other value- added services carried. Other intermediaries, more or less in competition with the provider potentially owing the duty, may assert rights to interconnections. The end user's access rights can be compared to those of a telephone subscriber. The supplier's access rights can be compared to those of a shipper by rail, truck, air, or sea seeking access to a transportation carrier, and to those of an originator of video programming seeking a distribution channel. The access rights of an intermediary seeking interconnections can be compared to those of a local bypass telephone carrier, to a non-wire line cellular telephone carrier seeking interconnections from a former Bell Company, or to a competing interexchange carrier like MCI seeking connections from AT&T.
The question of access rights implicates legitimate interests belonging both to the entity desiring access and to the entity refusing access. End users may seek access to receive content. Suppliers of content or value-added services may need access to reach particular markets in an economical fashion. For such entities, denial of access may be tantamount to exclusion from the market. Competing carriers have similar concerns. These concerns implicate public policy interests in competitive markets, as expressed by the antitrust laws, as well as the economic interests of the entity denied market access. For some entities, including content and value-added suppliers, end users, and competing carriers, denial of access may frustrate expectations based on representations made by suppliers of network services. Such frustration implicates the interests traditionally justifying contract enforcement.
Entities refusing access and opposing imposition of access duties may assert property rights in their systems and the privilege of free expression under the First Amendment. A core right in the bundle of rights that makes up property is the right to exclude. Free expression includes the privilege not to be forced to express disfavored ideas as well as the privilege to express favored ones. On this point, entities refusing access may conclude that access by a particular customer is inconsistent with the definition of that entity's product. For example, A may deny access to B because B wants to put content relating to the politics of health care reform onto A's discussion group limited to particular legal issues relating to the delivery of health care. Or, A may deny access to B because A fears that granting access to B will anger its other customers, resulting in a boycott. Alternatively, A may deny access in order to protect itself from potential liability for trafficking in certain kinds of pornography, for infringing on intellectual property rights, [FN14] or for invasion of privacy. [FN15] Viewed this way, the interests of suppliers seeking to limit access duties may be protected by the First Amendment and by the freedom of contract doctrine. [FN16] The privilege of using private property as the owner wishes, unless the government can justify interference under the takings doctrine, may also be implicated.
There is an additional public interest which may be involved, beyond the public interest in efficient markets and in the fulfillment of promise- based expectations. That is the public interest in quality infrastructure, exemplified by public support for roads and the post office, and by the strategic goal of universal service which underlies telephone regulation. The public needs an infrastructure-a system in which all the pieces work together- rather than merely a patchwork collection of largely independent proprietary communication services. Such a necessity justifies governmental intervention with respect to communications and information infrastructure because of the existence of network externalities. A consumer who connects to a communication service that is not connected to other communication services enjoys less utility than if interconnection exists. The increase in utility with the larger scope of interconnected networks is a network externality. [FN17] In certain circumstances, the presence of network externalities may serve as a sufficient economic incentive for private-sector decisionmakers to make the interconnections on their own. But the economic literature suggests that there is an important range of circumstances in which private decisionmaking will not lead to compatibility and interconnection, thus depriving consumers of the benefits of network externalities. [FN18] In other words, interconnected network effects *59 sometimes may be viewed as a public good [FN19] that suppliers have difficulty internalizing into their preference functions.
Figure 1
TABULAR OR GRAPHIC MATERIAL SET FORTH AT THIS POINT IS NOT DISPLAYABLE
TABULAR OR GRAPHIC MATERIAL SET FORTH AT THIS POINT IS NOT DISPLAYABLE
Figure 1 illustrates how the market, expectation-based, and public *60 interest dimensions have justified legal imposition of access duties. The X axis reflects holding out, the Y axis reflects monopoly power, and the Z axis reflects public interest. Common law and statutory common carriage doctrine has historically been justified both by the concepts of monopoly power and holding out. [FN20] Contract obligations, by comparison, have been justified predominantly by holding out alone. Finally, universal access and interconnection obligations have been justified largely by the public interest. The public interest dimension of Figure 1 would justify legal intervention aimed at ensuring interoperability even where the monopoly power and expectations dimensions would not justify such intervention.
III. THREE KINDS OF PROBLEMS
The positions, interests, and legal treatment of different entities in the infrastructure also can be thought of in terms of the relationships of such entities to each other and to the markets. Both vertical and horizontal relationships are important. In the vertical relationship, A supplies communications services and B supplies value-added features, such as finding and retrieval aids. A and B make a contract, giving B the exclusive right to supply a particular market. Competitors of B are excluded. In the horizontal configuration, A, B, and C compete in providing, for example, Internet connection services. A and B jointly operate a router and deny C access to the router, thus disadvantaging C in serving the market. Unilateral action also is conceivable. For instance, A may refuse to handle traffic from B because of political pressure from those opposed to B's point of view.
In order to make the analysis more concrete, this article assumes an NII architecture much like today's Internet, in which multiple backbones exist (such as NSFnet, CIX, or ANS CORE). [FN21] Under this architecture, multiple Internet access providers compete with each other to provide basic backbone connection services and also to offer certain value-added services (Cerfnet, Nearnet, Barnet, NetCom, and JVNCnet). A large number of Internet nodes [FN22] offer finding and retrieval aids like Gopher and World Wide Web. Other Internet nodes offer content through anonymous FTP or through integration with Web or Gopher servers. [FN23] Faced with increasing competition from competitive access providers in local markets, competing interexchange carriers in the interexchange market, *61 and traditional X25 public data networks like Sprintnet and BT Tymnet, interexchange and local access telephone companies offer switched digital and analog communications channels both on a dialup and dedicated-line basis.
It is important to understand that service providers in the NII range from pure communication service providers to nearly pure content providers. Between these extremes lie some of the most interesting new technology applications, including Gopher servers and World Wide Web servers. While the historical dichotomy between information providers and communications providers has collapsed under the pressure of new technology, appreciation of the continuum still is valuable.
IV. SOURCES OF RIGHT
There are a number of legal theories which have been advanced to assure access to the NII as possible alternatives or supplements to statutory common carriage and new statutory "open platform" requirements. [FN24] One such theory, still mainly speculative, involves imposing a duty based on common law common carrier obligations. This theory is built on common law concepts that antedate enactment of the Interstate Commerce Act-the first source of statutory common carriage. These common law concepts were meant to prohibit discriminatory denial of service by entities holding themselves out as serving everyone and possessing market power. [FN25] A second theory which may ensure access to the NII is antitrust essential facilities doctrine and related refusal-to-deal doctrine, which together prohibit denial of access to essential facilities by those in competition with the entities denied access. [FN26] The Supreme Court's decision in United States v. Terminal Railroad Ass'n [FN27] sets forth the basic standard for requiring an entity controlling an essential facility to provide access to nonparticipating competitors. [FN28]
Probably the most attractive doctrine, because of its flexibility, is that of express and implied contract. Under this approach, a service provider that holds itself out as serving the public would be prohibited from denying access without proper justification. Contract approaches are limited *62 by the historical treatment of credit card arrangements and of commercial advertisements in which the representation by the provider is treated, not as an offer, but rather as a solicitation of offers. [FN29] The efficacy of a contract-based regime for assuring access could be strengthened by an approach to tort immunity that extends immunity only to those providers that publish, possibly through some central database, their terms of service and disavow any intent to screen or censor content. [FN30]
A. Statutory Common Carrier
The best known source of access rights and duties in communications networks is the set of statutory common carrier obligations imposed under title II of the Federal Communications Act of 1934 (Title II). [FN31] These statutory provisions extend a statutory common carrier scheme previously outlined by the Interstate Commerce Act, [FN32] which in turn was derived with little change from common law common carrier obligations. [FN33] Under Title II, common carriers must furnish communication services upon reasonable request, furnish physical connections with other carriers, establish "through routes," charges, and allocation schemes for the charges among participating carriers, and establish facilities and regulations for through routes. [FN34] Carriers may not discriminate unjustly or unreasonably or give undue or unreasonable preferences to different classes of users. [FN35] The charges, practices, classifications, and regulations of such carriers must be just and reasonable. [FN36]
Of course, nondiscrimination and interconnection are not the only obligations imposed on common carriers under title II of the FCA. The fact that other duties may exist is part of the problem with pursuing access rights under Title II. Moreover, the Supreme Court's decision in MCI Telecommunications Corp. v. AT&T [FN37] raises questions about the FCC's power to simplify the details of common carrier regulation. [FN38]
*63 1. Interconnection requirements
The requirement to maintain physical interconnections and through routes is derived from section 1(4) of the original Interstate Commerce Act. [FN39] The Interstate Commerce Act's original obligation provisions were expanded by the Mann-Elkins Act of 1910 [FN40] to eliminate a burdensome precondition to administrative action imposed by ICC v. Northern Pacific Railway. [FN41] That case involved an ICC order requiring the Northern Pacific Railway to offer through routes and joint rates for passengers and their baggage to and from points on the Chicago and Northwestern Railway between Illinois and Iowa, as well as points on the Union Pacific Railroad between Colorado, Nebraska, and Missouri. [FN42] The joint rates were to be the same as the present rates between the same points via the Northern Pacific. [FN43] All of the points were reachable by the Northern Pacific or its connecting railroads. [FN44] The Supreme Court, however, held that the ICC had no authority to issue such an order in the absence of a preliminary finding that no reasonable or satisfactory through route already existed. [FN45] Almost immediately after this holding was issued, Congress enacted the Mann-Elkins Act provision eliminating the precondition of making that finding. [FN46]
In the communications context, a number of early disputes resulted from the Bell System's policy of refusing to connect with certain independent telephone companies between 1894 to 1906. Thirty-four states enacted laws compelling physical connection. [FN47] Addressing the phone *64 system context, the interconnection obligation under the Federal Communications Act is a reasonableness obligation. [FN48] The most significant interconnection decision was the FCC's Carterfone [FN49] decision in 1968, which started the unraveling of the Bell System. [FN50] The Carterfone decision has been narrowly interpreted by subsequent court decisions only to prohibit telephone company refusals of "harmless" interconnections, [FN51] while allowing refusals based on justifications of detriment to the public or adverse effects on the telephone system. [FN52] In rationalizing regulation of "record carriers," Congress required then-dominant Western Union and RCA to provide unrestricted interconnection to the lines of other record carriers, [FN53] while exempting carriers that did not control significant market shares from certain interconnection requirements. [FN54] Yet, the distinction between dominant and nondominant carriers in application of common carriage law to telephone carriers was essentially rejected by the Supreme Court in MCI Telecommunications Corp. v. AT&T . [FN55] While Congress had explicitly sanctioned the distinction with respect to record carriers, it had remained silent with respect to telephone carriers.
*65 Another more recent interconnection controversy arose over access by nonwire line paging services to local access telephone carriers. [FN56] The telephone companies involved in this controversy were concerned that interconnections would be noncompensatory because of short holding times for beeper calls. [FN57]
The effect of the interconnection decisions is threefold. First, the decisions do not recognize a general common law obligation to allow competing services to interconnect. Second, however, the holdings impose a burden under section 201 on common carriers to justify their interconnection refusals. Finally, the courts may recognize legitimate justifications for refusal to interconnect based either on technical harm or noncompensatory rates, in light of the potential high costs of certain intercon- nections.
In the background is the idea that interconnection obligations are easier to rationalize for dominant carriers than for nondominant ones, at least when Congress articulates deferential treatment for nondominant carriers. [FN58] Combined with the recent case law invalidating on-premises interconnection requirements, [FN59] it may be that interconnection requirements can be justified constitutionally only on a threshold showing of the existence of monopoly control over a network facility and the absence of acceptable alternatives.
2. Modification of the common carrier category
By the early 1970s, it had become evident that the administrative apparatus erected to apply common carrier requirements was imposing unacceptable costs on the evolution of the communications infrastructure. [FN60] Accordingly, the FCC began to modify its interpretation of the common carrier category, narrowing it to allow more services to be offered free of the restraints imposed by common carrier obligations. [FN61] The FCC also exempted certain nondominant common carriers from most of the detailed tariff requirements. [FN62] When the FCC decided to allow competition *66 in the specialized common carrier market, it did so in large part by requiring the telephone companies to allow entities such as MCI and Southern Pacific Communications to interconnect. [FN63]
However, in AT&T v. FCC , [FN64] the District of Columbia Circuit held that the FCC's order exempting nondominant interexchange carriers such as MCI from tariff requirements violated the Communications Act. [FN65] Once common carrier status was found to exist, the court held, the FCC lacked discretion to tailor the requirements of Title II. [FN66] The basic holding of this case eventually was affirmed by the Supreme Court in MCI Telecommunications Corp. v. AT&T . [FN67] The effect of this precedent is to limit the FCC's flexibility to adapt the existing Title II common carrier obligations to the particular needs of a digital information infrastructure like the NII. Once common carrier status is found to exist, the traditional statutory requirements may rigidly apply.
B. Narrowed Statutory Duty
The combination of dissatisfaction with the traditional common carrier requirements and a continued perception that access entitlements are necessary in some parts of the information infrastructure has led Congress to impose more specialized access requirements, most notably in cable television regulation. The Cable Communications Policy Act of 1994, as subsequently amended, [FN68] obligates cable television operators to provide access to public, educational, and governmental programming, [FN69] and to unaffiliated commercial programming. [FN70] In addition, cable operators must carry local television station programming. [FN71] However, these entitlements were called into question in Turner Broadcasting Systems, Inc. v. FCC , [FN72] decided by the Supreme Court in 1994.
*67 More recently, during the congressional debate over new communications legislation, some authorities have begun to view with sympathy proposals for open platform obligations which would essentially require interoperability.
C. Common Law Common Carrier
Statutory common carrier obligations are derived, with little change, from common law common carrier obligations. [FN73] At common law, someone engaged in a public calling, like an innkeeper or a ferry operator, was treated as a common carrier. [FN74] The early cases are somewhat confused in their specification of the determinants of common carrier status, but it seems evident that monopoly power, "holding out" (a promise, in essence, to serve all comers), and public interest in having the service performed affordably and fairly were the three important determinants. [FN75]
Once a business was classified as a common carrier, the original obligation imposed by law was one of nondiscrimination. [FN76] Any refusal to serve had to be justified by a legitimate economic interest. [FN77] Gradually, as alleged discrimination took the form of unreasonable charges, courts applying common carrier duties began to scrutinize pricing. [FN78] But the common law duty not to discriminate apparently extended only to end users, and did not include a requirement to provide interconnections with competitors. [FN79]
The development of common law common carriage doctrine was arrested by the enactment of the Interstate Commerce Act in 1889, [FN80] the Federal Communications Act in 1934, [FN81] and similar state statutes. These statutes largely preempted the operation of common law with respect to common carriers. Now, however, with deregulation and retraction of the scope of statutory common carriage, there may be a renewed opportunity for the development of common law concepts, although cases addressing this topic have not yet arisen.
*68 D. Contract
Contract rights to access may be based on representations made by the service provider to induce people to become customers or may be based on representations and conduct once a provider-customer relationship has been established. Implied contract doctrine, more than any statutory or other common law theory, appeared to undergird the recent claims of an Arizona law firm that it should not be disconnected from the Internet after it flooded Internet newsgroups with advertisements in 1994. [FN82] The implied contract theory has a major advantage for persons claiming access rights, in that it does not require them to make any showing of the absence of alternative sources of supply. [FN83] Implied contract also offers certain advantages to the entity denying access, in that such an entity may negate potential contract claims with appropriate disclaimer language.
There are several branches of contract law that are important in the context of information infrastructure access rights. First, the relationship between the service provider and users of the network must be evaluated as a bargained-for exchange. [FN84] In the usual situation, a service provider unilaterally announces its terms, probably without knowing the identities of the people who may use the network and certainly without any of the give-and- take that most people associate with bargaining. Persons desiring access to the service simply use the network after learning of the service terms. Such use may be viewed as an acceptance of the offer represented by the published terms of service. Alternatively, the consumer's use of the service may be deemed the offer itself, responding to a solicitation of offers represented by the publication of terms. Then the provider's acceptance of the offer must be identified.
A second applicable theory, an alternative to bargained-for exchange, is promissory estoppel. Under section 90 of the Restatement (Second) of Contracts, [FN85] the person making a promise may be bound to perform that promise if someone else reasonably relies on that promise to his detriment, and it was reasonable for the promisor to anticipate reliance under the circumstances. [FN86] The detrimental reliance element might or might not be present in the typical network transaction. Yet, it is conceivable that in some situations, passing up other network subscription opportunities or going to the trouble to arrange telecommunication scripts and to distribute information about e-mail addresses on the basis of representations made by a particular provider may constitute reliance.
The bargain theory is potentially limited by advertising cases. Terms *69 of access communicated to the general public resemble advertisements, and advertisements have not generally been considered to make the publishers thereof liable for lack of compliance with their terms. [FN87] Statements in advertisements traditionally were construed as solicitations of offers rather than offers themselves. [FN88] Thus, a refusal to sell on the terms communicated in the advertisement did not breach a contract; it merely was a rejection of the buyer's offer to make a contract. [FN89] However, this general rule was not applied when an advertisement manifested a clear intent to make a promise. For example, a statement of definite price, accompanied by the phrase "first come, first served," may amount to an offer. [FN90] The most famous case in this area is Carlill v. Carbolic Smoke Ball Co. , [FN91] in which an advertiser promised to pay a 100 reward to anyone who contracted a cold after using the advertised smoke ball. [FN92] Evidencing the promissory intent, the advertisement also stated that the advertiser had deposited 100s with a bank as, in effect, an escrow agent. [FN93] The court found that the nature of the communication evidenced an intent to make an offer notwithstanding the risk of attracting a large number of offerees. [FN94]
The Restatement of Contractsharmonizes this caselaw by establishing a presumption that advertisements are ordinarily intended as mere solicitations of offers rather than as offers. However, the Restatement acknowledges, first, that one may make an offer through an advertisement, [FN95] and second, that an advertisement that is not an offer nevertheless may contain promises or representations that become part of the eventual contract. [FN96] In the network access context, the advertisement cases could be applied to allow the network services provider to refuse any particular request for service, but bind the provider to deliver service according to its published terms once a request for service is accepted. [FN97]
*70 Further, terms of access published by a network services provider also are not unlike the terms published by the issuer of a credit card. Such terms are intended to reach a very large number of people, who will subsequently enter into discrete transactions, presumably relying on the published terms. The prevailing view is that credit card terms do not give rise to enforceable obligations to allow access to the credit represented by the card. [FN98] Rather, the terms are revocable offers which are accepted each time the card holder uses the card. [FN99] "The credit card relationship, properly analyzed, should be viewed as an offer by the issuer to create the opportunity for a series of unilateral contracts which are actually formed when the holder uses the credit card to buy goods or services or to obtain cash." [FN100]
If applied strictly to the network access context, the credit card cases would permit the service provider to change the terms of the contract virtually continuously. Each request for service would be a revokable offer for a new contract. If the network provided service on that occasion, a contract would be formed only for the duration of that particular transaction on the network. Because no lasting contract would be in existence, the service provider would remain free to disconnect service at almost any time, so long as the disconnection occurred between the end of one transaction and the commencement of the next.
Both the advertising and credit card cases, if applied in the network access context, would leave the service provider free to deny service until a particular request for service was accepted. To understand the implications of this approach, it is crucial to understand what constitutes a "request*71 for service" in the NII context. If each connection request in the TCP protocol constitutes a separate request for service, the obligations of the network service provider would remain in effect only until it delivered a particular file or menu item or pointer in response to a World Wide Web or Gopher query, or until the termination of a particular FTP or Telnet session. [FN101] This characterization of "request for service" would impose little obligation on the service provider and would confer on the requester legal rights of minimal practical utility. [FN102]
A somewhat greater range of obligations and entitlements would result from characterizing a request for service as a request to use a system for a particular billing period, assuming that billing takes place on approximately a monthly cycle. But even if the thirty days of rights or obligations are of practical significance, it is not likely that the NII will function effectively on thirty-day billing cycles. It would be much more advantageous for billing cycles to be shorter, considering the granularity of billing, and for billing thus to occur as it does between long-distance service providers and individual telephone subscribers-on a per-call basis, albeit aggregated through the monthly billing by local telephone service provider to subscriber.
In general, the implied contract theory provides flexibility for market- based definitions of infrastructure. Even when the classical metaphor of two equal parties sitting across a table and explicitly bargaining about their contract terms is unhelpful because of its remoteness from reality, the freedom afforded both offeror and offeree in the context where the offeror unilaterally specifies in some detail its terms of service and reserves powers to amend or terminate is considerable and makes the implied contract theory attractive. The offeree knows exactly what she is getting into before making an economic commitment. Both bargain theory and promissory estoppel theory give the offeror the power to limit his liability. An offeror can specify how acceptance is to be made, [FN103] and may define not only the duration of any contract to be formed by acceptance, [FN104] but also how long the offer will remain in effect. [FN105] Promissory estoppel theory creates enforceable promises only when detrimental reliance is reasonable. [FN106] Thus, to limit liability exposure, the offeror may circumscribe the range of reasonableness and reliance by adjusting the content of his statements.
Nevertheless, these theoretical advantages of implied contract doctrine *72 do not make it preferable to the other legal doctrines unless contracts really would be enforceable under the circumstances likely to exist in the market for infrastructure services. If the line of cases negating contractual entitlements for credit card subscribers [FN107] is used to make commitments by network services providers similarly unenforceable, then contract theory becomes irrelevant as a policy matter, necessitating greater reliance on other theories.
Before moving beyond contract theory, it is useful to make certain further observations. The availability of competitive alternatives to a particular facility or service enters into contract analysis by affecting detrimental reliance. If there are many alternatives to a particular provider's service, and the costs of switching are trivial, then someone who intends to use the service in reliance on the provider's statement of terms has suffered little in the way of detrimental reliance if service on those terms is denied. On the other hand, if there are few alternatives or if switching costs are high, the cost of disappointed expectations is greater, increasing the likelihood of enforcement of the promise on a detrimental reliance basis, and increasing the likelihood of damages under the bargain theory.
Contract doctrine may be attractive to suppliers of network services because it gives them greater power to define and limit their liability. By the same token, contract theory may be unattractive to consumers of network services if they lack the power to bargain for terms that protect them. Lack of power may come from disparate size or wealth, or it may derive from the transaction costs of bargaining. If a consumer of network services only desires one second of access to a World Wide Web server, the cost of off-line haggling over contract terms is too high to be borne. On the other hand, if transaction costs of bargaining can be managed, then contract theory can be an attractive regime for consumers of network services because it gives them the power to decide what bargain best meets their needs. The adverse effects of high transaction costs, like disparate bargaining power, can be mitigated by collective action. Groups of consumers of network services and suppliers of network services can negotiate standard terms of service, and then those terms can be incorporated by reference into small transactions. Antitrust concerns may limit collective actions, however.
The distributed nature of the NII weakens the distinction between suppliers and consumers of network services, since everyone may be both a supplier and a consumer at various times. [FN108] Thus any ideological preference for supplier interests or consumer interests tends to become irrelevant. The NII era may not be one dominated by a consumerism verses business struggle; rather, it may be one in which efficient commerce coincides with consumer welfare.
Alternatively, the effects of high transaction costs and disparate bargaining power could be mitigated by the legislative process, resulting in a *73 statute extending the binding effect of published terms of service. This is essentially what has been done with respect to credit cards. [FN109] In the network access context, a statute could require network service providers to post their terms of service on a publicly accessible database.
It is worth noting that the concept of a publicly accessible database of network service terms, backed up by enforceability of the terms as contracts, shares certain key features with the traditional idea of common-carrier tariffs. [FN110] The published terms are extended to anyone in the targeted market, thus discouraging discrimination among similarly situated competitors. Additionally, the published terms crystallize the terms of the contract. Tariff terms are enforceable, but other arguable contract terms not contained in the tariff are unenforceable. [FN111] In summary, there are six prerequisites to a meaningful contract regime in the NII context. First, it must be clear which representations constitute enforceable promises and which ones are mere nonpromissory representations of fact. Second, the promissory representations must be treated as such and not treated as mere invitations to make offers. Third, the statute of frauds must not prevent enforcement of contracts made by conduct after publication of the contract terms. Fourth, publication of contract terms should suffice without the offeree having to prove subjective knowledge and detrimental reliance. Fifth, and perhaps most challenging, disclaimers of enforceability and reservations of power to amend or terminate access must be circumscribed so they do not swallow up the affirmative theories of contract enforcement. Finally, the effect of competitive alternatives should be defined.
E. Antitrust
Under current interpretations of antitrust doctrine, [FN112] the probability of antitrust liability for NII providers is low because all of the established and reasonably anticipated markets for infrastructure services are competitive and have low barriers to entry. Nevertheless, certain joint arrangements for packet routing and directory services, as well as tying *74 arrangements bundling different network and information services, may invite close antitrust scrutiny. Additionally, the analytical framework for all of the relevant antitrust doctrines involves a scrutiny of market structures similar to that which enters into any serious analysis of common carrier status or policy-based rationales for imposing new access obligations. [FN113]
1. Joint ventures
A joint venture has been defined for antitrust purposes as including the following attributes: (1) the enterprise is under the joint control of the parent firms, which themselves are not under related control; (2) each parent makes substantial contribution to the joint enterprise; (3) the enterprise exists as a business entity separate from its parents; and (4) the joint venture creates significant new enterprise capability in terms of new productive capacity, new technology, a new product, or entry into a new market. [FN114]
Joint ventures are essentially partnerships to accomplish a particular task or project. In the NII context, joint ventures could be useful structures for providing certain common services, for example, routers to exchange traffic among network service providers. [FN115] Joint ventures potentially violate the antitrust laws in that they produce goods or services *75 that otherwise would be produced by individual partners in competition with each other. Joint ventures thus amount to agreements to restrict output, because they implicitly involve each partner withholding its own unilateral efforts to produce the good or service that is to be produced by the joint venture. Joint ventures are evaluated under the rule of reason. [FN116] When the joint venture partners sell mostly in exclusive geographic territories, the likelihood of the joint venture violating the antitrust laws is much lower because any negative effect on output is reduced. [FN117] Generally, joint ventures whose procompetitive effects outweigh their anticompetitive effects are permissible under section 1 of the Sherman Act. [FN118] Thus, Broadcast Music, Inc. v. CBS [FN119] approved a copyright collective that used standard license forms for copyrighted music even though the license included a standard fee. [FN120]
On September 15, 1993, the Justice Department issued an enforcement policy statement for joint ventures in the health care industry. [FN121] The policy statement provides for "antitrust safety zones" which immunize certain types of joint ventures. [FN122] Ventures involving high technology or other expensive equipment, the cost of which must be shared, would not be challenged. [FN123] The policy statement thus would protect arrangements such as joint ventures among rural hospitals to share MRIs or agreements among community hospitals jointly to operate helicopter services. [FN124] The Justice Department's statement further suggested a four-step rule of reason analysis for joint ventures: (1) define the relevant market; (2) evaluate the competitive effects, particularly focusing on the possibility that the joint venture would eliminate an existing or potentially *76 viable competing provider of the service; (3) evaluate the potential procompetitive efficiencies of the venture; and (4) evaluate ancillary agreements or conditions that might unreasonably restrict competition and would be unlikely to contribute significantly to the legitimate purposes of the joint venture. [FN125] While the policy statement focuses on health care joint ventures, its analytical approach may also be applicable to joint ventures in the NII.
A recent law review article supplements the Justice Department framework for evaluating joint ventures, specifically noting the trend toward joint ventures in the telecommunications industry. [FN126] The author suggests beginning with an assessment of the restriction on competition, acknowledging that " a n inevitable competitive loss occurs when parties who are rivals in a particular area suspend that rivalry in order to cooperate with each other." [FN127] The next step would be to evaluate efficiencies generated by the joint venture, [FN128] including reduction of risk, [FN129] economies of scale, [FN130] elimination of wasteful redundancies, [FN131] access to complimentary resources, [FN132] and making unique products available to consumers. [FN133] The article further proposes evaluating joint ventures along a continuum reflecting the degree to which the parties have combined their resources, scrutinizing highly integrated joint ventures more rigorously than unintegrated arrangements. [FN134]
2. Vertical exclusive arrangements
In the paradigmatic vertical relationship, A supplies communications services and B supplies value-added features, such as finding and retrieval aids. A and B make a contract, giving B the exclusive right to supply a particular market. Competitors of B get excluded from the downstream market for A's product, and competitors of A get excluded from B's downstream services. Courts analyze vertical exclusive arrangements more sympathetically under section 1 of the Sherman Act than they do horizontal exclusive arrangements. [FN135] This is because vertical arrangements *77 frequently strengthen interbrand competition, (competition between A and its competitors) even though they may limit intrabrand competition (competition between B and its competitors for distribution of A's product). [FN136] For example, an exclusive arrangement between a legal database provider and an Internet access provider in a particular geographic area might strengthen the ability of both partners to compete against other Internet access providers and other database providers. However, the arrangement would reduce potential intrabrand competition that would occur if multiple Internet access providers competed with each other for customers desiring access to the database.
Futurevision Cable Systems of Wiggins, Inc. v. Multivision Cable TV Corp. [FN137] involved an unsuccessful claim by a new entrant cable network that exclusive contracts between ESPN, The Learning Channel, and existing cable networks violated section 1 of the Sherman Act. The federal district court concluded that the plaintiff failed to show market power on the part of ESPN or The Learning Channel. [FN138] The plaintiff was unable to demonstrate that the exclusive arrangement prevented it from entering the market and competing vigorously with other sources of programming. [FN139] This case is interesting because it involved exclusive arrangements between content providers (ESPN and The Learning Channel) and a network services provider (the existing cable channels) in which the existence of alternatives (other sources of programming service) nullified an antitrust claim brought by a new entrant in one part of the market. The fact pattern is roughly equivalent to that of a new network services provider challenging exclusive arrangements between content providers and existing network services providers in the Internet context.
3. Tying arrangements
Tying occurs when a seller bundles components of a system and refuses to sell them separately. An information services provider might bundle content with tools for indexing, searching, and retrieval, such as World Wide Web, and then refuse access to the content unless one also buys access to the provider's Web server.
The essential characteristic of an invalid tying arrangement lies in the seller's exploitation of its control over the tying product [the content] to force the buyer into the purchase of a tied product [the Web server] that the buyer either did not want at all, or might have preferred to purchase elsewhere on different terms. When such "forcing" is present, competition on the merits in the market for the tied item is restrained and the Sherman Act is violated. [FN140]
*78 In Eastman Kodak Co. v. Image Technical Services, Inc. , [FN141] the Supreme Court analyzed the impact of antitrust tying doctrine on product bundles. The Court rejected two proposed limitations on defining the relevant market which would have made it harder to find an illegal tying arrangement. First, the Court rejected the argument that functionally linked products, one of which is useless without the other, can never produce liability if used in a tying arrangement. [FN142] Rather, the Court held that for two functionally linked products to be considered two distinct products for purposes of tying liability, "there must be sufficient consumer demand so that it is efficient for a firm to provide them separately ." [FN143] On the facts, the Eastman Kodak Court found evidence that service and parts had been sold separately in the past and noted that the very fact that a service industry had developed, members of which had challenged Kodak, was itself evidence of the efficiency of a separate market for service. [FN144]
In the digital network services area, an analogous example would consist of facts evidencing that it is efficient to sell content separate from Web service, or that it is efficient to produce Gopher and World Wide Web servers and sell access to them independently of Internet access. Also, an analogy would exist if facts showed that it is efficient to produce and sell chunking and tagging value separate from pointers value. [FN145]
Second, the Supreme Court rejected Kodak's contention that, as a matter of law, a single brand of a product or service can never be a relevant market under the Sherman Act. [FN146] The Court stated:
The relevant market for antitrust purposes is determined by the choices available to Kodak equipment owners. Because service and parts for Kodak equipment are not interchangeable with other manufacturers' service and parts, the relevant market from the Kodak-equipment owner's perspective is composed of only those companies that service Kodak machines. [FN147]
Similarly, the market for PC Internet interface software for CompuServe could be a relevant market for purposes of NII antitrust analysis.
*79 Having thus defined the market, the Eastman Kodak Court proceeded to assess Kodak's behavior. The Court noted that while a firm can refuse to deal with its competitors, it must offer legitimate competitive reasons for the refusal, other than merely wishing to drive them from the marketplace. [FN148] The facts showed that Kodak declined to sell parts to its equipment customers unless they also bought its service or repaired their own equipment. [FN149] On the issue of Kodak's behavior, independent service organizations, as the plaintiffs in the antitrust action, argued that Kodak's restrictive repair parts policy hurt their business and violated the antitrust law. [FN150] In affirming the Ninth Circuit's refusal to affirm summary judgment for Kodak and in remanding the case for trial, the Court left the door open for the plaintiffs to prevail. [FN151]
The dissent, which would have reversed the Ninth Circuit's decision, characterized the majority opinion as giving the wrong answer to the question of
[w]hether, for purposes of applying the per se rule condemning tying arrangements, and for purposes of applying the exacting rules governing the behavior of would-be monopolists, a manufacturer's conceded lack of power in the interbrand market for its equipment is somehow consistent with its possession of "market," or even "monopoly," power in wholly derivative aftermarkets for that equipment. [FN152]
Because of its final posture, remanding on a "sparse" record [FN153] for development of factual evidence on the crucial predicates, the Eastman Kodak opinion is in the end only an invitation to speculate about the role of antitrust tying doctrine in ensuring access to aspects of the NII. Nevertheless, the case identifies one theory for forcing a producer with substantial market power to unbundle: the producer's failure to justify continued bundling on cost or other grounds other than excluding competitors. [FN154]
4. Exclusive dealing
An exclusive dealing arrangement [FN155] is potentially violative of the antitrust laws because it may reduce competition in either the upstream or downstream market.
*80 Exclusive dealing can have adverse economic consequences by allowing one supplier of goods or services unreasonably to deprive other suppliers of a market for their goods, or by allowing one buyer of goods unreasonably to deprive other buyers of a needed source of supply. In determining whether an exclusive-dealing contract is unreasonable, the proper focus is on the structure of the market for the product or services in question-the number of sellers and buyers in the market, the volume of their business, and the ease with which buyers andsellers can redirect their purchases or sales to others. Exclusive dealing is an unreasonable restraint on trade only when a significant fraction of buyers or sellers are frozen out of a market by the exclusive deal. [FN156]
Exclusive dealing contracts can violate both section 3 of the Clayton Act [FN157] and section 1 of the Sherman Act. [FN158] "The legality of an exclusive dealing arrangement under the Clayton Act depends on whether the competition foreclosed constitutes a substantial share of the relevant market." [FN159] The absolute percentage of potential market foreclosure is not the sole issue, however. One also must consider the probable effect of the contract on competition, taking into account the relative strength of the parties, the probable immediate and future effects which preemption of that share of the market might have on effective competition, and the existence of legitimate business justifications for the contracts. [FN160]
For example, suppose Cerfnet, a midlevel network in the Internet, enters into an exclusive dealing arrangement with the Cleveland Freenet, which provides local Internet access services. This arrangement arguably would reduce competition in the local Internet access services market because no other provider of Internet access services would be able to deal with Cerfnet. It also could be said to reduce competition in the market for midlevel Internet services because no other midlevel network would be free to deal with Cleveland Freenet. But in order for such an arrangement to violate the antitrust laws, there would have to be a showing of (1) a reduction of competition, and (2) the absence of procompetitive justification. [FN161] If the markets for both midlevel network services and network access services are competitive (they are), and if neither party to the exclusive dealing arrangement possesses significant market power (they do not), then the arrangement is permissible. [FN162]
*81 In addition, even if competition in either or both markets were reduced, the arrangement still might not lead to antitrust liability if it could be justified on the grounds of improving the market position of either Cerfnet or Cleveland Freenet, thus enhancing interbrand competition. The possibility of new entry, even when there are significant barriers to entry, also reduces the likelihood of a successful claim on a theory in which market power is an element. [FN163] Thus, for example, in a hypothetical claim against West Publishing Company for an exclusive dealing arrangement with an Internet services provider, West could probably resist a showing of market power in the market for remotely accessible comprehensive legal databases by demonstrating a high potential for entry by new competitors into that market. [FN164]
5. Concerted refusals to deal
Concerted refusals to deal are group boycotts and may constitute per se violations of the Sherman Act. [FN165] Concerted refusal to deal claims frequently arise in the context of joint ventures because it is logical for joint venture partners to exclude nonparticipants from the benefits of the venture. Recent caselaw indicates the outer limits of the refusal to deal doctrine.
In SCFC ILC, Inc. v. VISA USA, Inc. , [FN166] a district court approved a jury determination that a decision by the VISA credit card joint venture to exclude Sears Roebuck affiliates from VISA membership because of Sears' competing Discover Card constituted a violation of section 1 of the Sherman Act. The district court rejected the argument that the affiliates had to show that the VISA joint venture was an essential facility in order to find liability. [FN167] Instead, the court stated that joint ventures ordinarily *82 are evaluated under the rule of reason analysis, [FN168] which asks whether "the challenge to agreement is one that promotes competition or one that suppresses competition." [FN169] The SCFC court rejected VISA's reliance on two legal "screens" that would have applied a rule of reason analysis to dismiss the case before it reached the jury. The first screen involved a finding of a lack of market power; the second involves a finding that the claims of the plaintiff make no economic sense in terms of weighing the anticompetitive and procompetitive effects of the allegedly harmful conduct. [FN170] The district court also rejected VISA's argument that joint ventures are entitled to more deferential section 1 analysis than conduct involving independent entities because of the procompetitive effect of joint ventures. [FN171]
In rejecting VISA's arguments, the district court found sufficient evidence to support a jury conclusion that VISA Bylaw 2.06 had substantial harmful effects on competition, based on evidence that exclusion from highly profitable VISA membership would be a strong disincentive for anyone to introduce a new card if that would disqualify it from membership in VISA. [FN172] The court also approved the jury's rejection of VISA's argument that the joint venture had a beneficial procompetitive effect by enhancing competition at the intersystem level. [FN173] It is important to understand that the district court did not reject this procompetitive effects argument as a matter of law; rather, the district court simply allowed the jury to reject the argument as a matter of fact.
However, the Tenth Circuit Court of Appeals reversed the district court on the basis of a rule of reason analysis which focused on the level of market power involved in the case and on VISA's efficiency justification for Bylaw 2.06. [FN174] First, the circuit court held, as a matter of law, that the evidence was insufficient to show that VISA had market power, thus applying the first "screen" which the lower court had rejected. [FN175] The circuit court's analysis supporting this conclusion stressed that the bylaw-which Sears alleged to have harmed competition-was not, in itself, the proper focus of the market power analysis. [FN176] Rather, it was the bylaw's effects on the market which were determinative. The court found no evidence that the bylaw increased prices, reduced output, or other anticompetitive effects outweighing VISA's procompetitive rationales for *83 the bylaw. [FN177] Thus, Bylaw 2.06 did not alter the character of the market or harm consumers. [FN178]
The Tenth Circuit court proceeded to analyze VISA's procompetitive justification for Bylaw 2.06. The court accepted VISA's argument that the bylaw was implemented to protect VISA's property from competitors who otherwise would enjoy a free ride. [FN179] Since the effect of such free riding would be to increase costs to VISA members unfairly, VISA's response to free riding in the form of the bylaw was permissible. [FN180]
6. Essential facilities
a. In general . The essential facilities doctrine in antitrust law imposes liability for money damages and possible injunctive relief upon a monopolist or a group of competitors sharing monopoly power who refuse to allow a competitor access to an essential facility. [FN181] It is relatively rare for a plaintiff to meet the requirements for a showing of essential facilities liability. The leading cases in this area show the diversity of factual contexts implicated. These cases have involved (1) a group of railroads that formed a partnership to operate the only terminal railroad company in St. Louis, access to which was practically necessary in order to exchange traffic through the St. Louis gateway; [FN182] (2) denial to MCI by AT&T of access to its long distance network; [FN183] and (3) denial of access to an electric power grid to customers who purchased power from other sources. [FN184]
Four elements must be shown to trigger essential facilities liability: (1) control of an essential facility by a monopolist, (2) a competitor's inability practically or reasonably to duplicate the essential facility, (3) denial of the use of the facility to the competitor, and (4) feasibility of providing access to the facility. [FN185] These elements usually are difficult to establish on facts concerning digital information networks because the *84 availability of a wide range of alternative paths between any two points tends to negate the first two elements. [FN186]
The Supreme Court's decision in United States v. Terminal Railroad Ass'n [FN187] established the basic standard for when access for nonparticipating competitors would be required when competitors control an essential facility. The case arose after several railroads entering St. Louis organized a terminal company, which, by acquiring two competitors, obtained a monopoly in the market for interchange of traffic in the St. Louis terminal. [FN188] The Supreme Court recognized that jointly-owned terminal facilities could enhance trade. [FN189] Moreover, the Court found that
in ordinary circumstances, a number of independent companies might combine for the purpose of controlling or acquiring terminals for their common but exclusive use. In such cases other companies might be admitted upon terms or excluded altogether. If such terms were too onerous, there would ordinarily remain the right and power to construct their own terminals. [FN190]
Thus, the Court recognized the significant role that the availability of alternatives plays in applying the antitrust laws to joint ventures controlling essential facilities.
However, on the facts of the case, physical and topographical conditions prevented the construction of alternative means of access. [FN191] In such circumstances, "a unified system is an obstacle, a hindrance, and a restriction upon interstate commerce, unless it is the impartial agent of all who, owing to conditions, are under such compulsion, as here exists, to use its facilities." [FN192] While there was no showing that the terminal company had excluded nonparticipating carriers, there was no guarantee that it would not do so in the future. [FN193] Moreover, the company imposed certain rates that disadvantaged nonparticipating carriers. [FN194] To remedy this situation, the government urged that the company be dissolved. [FN195] The Supreme Court, however, viewed dissolution as an excessive remedy, and instead remanded the case for entry of a decree directing the parties to submit a plan for the reorganization of the contracts among the companies participating in ownership and operation of the company to effect the following goals:
*85 First. By providing for the admission of any existing or future railroad to joint ownership and control of the combined terminal properties, upon such just and reasonable terms as shall place such applying company upon a plane of equality in respectof benefits and burdens with the present proprietary companies.
Second. Such plan of reorganization must also provide definitely for the use of the terminal facilities by any other railroad not electing to become a joint owner, upon such just and reasonable terms and regulations as will, in respect of use, character, and cost of service, place every such company upon as nearly an equal plane as may be with respect to expenses and charges as that occupied by the proprietary companies.
Third. By eliminating from the present agreement between the terminal company and the proprietary companies any provision which restricts any such company to the use of the facilities of the Terminal Company. [FN196]
In MCI Communications Corp. v. AT&T , [FN197] the United States Court of Appeals for the Seventh Circuit affirmed, in material part, a judgment of antitrust liability against AT&T for denying interconnections to MCI. The court found that AT&T controlled essential facilities and that the evidence supported the jury's determination that AT&T denied the essential facilities (interconnections for FX and CCSA service) when access to such facilities could have been feasibly provided. [FN198] The fact that MCI, the entity desiring interconnection, had not actually built the facilities to make use of the interconnection did not defeat its claim. [FN199] On the other hand, the court distinguished MCI's claim for multipoint connections. [FN200] The court rejected MCI's antitrust challenge to AT&T for refusing to provide multipoint interconnections that would have given MCI access to AT&T's entire long distance network. [FN201] Rather, the court concluded that AT&T's refusal to assume voluntarily "the extraordinary obligation to fill in the gaps in its competitor's network" did not suffice to support a finding that it was trying to maintain its monopoly of long distance telephone service by anticompetitive means. [FN202]
Interface Group, Inc. v. Gordon Publications, Inc. [FN203] involved an unsuccessful essential facilities claim by a trade magazine excluded from a computer trade show. The district court distinguished Terminal Railroad on the grounds that the trade show was not an essential facility, because the market included a number of other channels for the plaintiff to communicate *86 its advertising. [FN204] The analogies between these facts and the facts in the Internet examples are useful. Advertising space resembles bandwidth that might be sold by a value-added network services provider denied bit services by a conduit. [FN205] The journal in Gordon Publications is analogous to the network services provider, while the trade show is analogous to the conduit.
Also of interest in the context of infrastructure access is Illinois ex rel. Burris v. Panhandle Eastern Pipeline Co. [FN206] Historically, natural gas pipelines bought gas at the wellhead, transported it, and resold it to customers of the pipeline, thus bundling the gas with its pipeline transportation. [FN207] Panhandle purchased gas at high prices, and wellhead prices for gas subsequently fell. [FN208] Certain customers of Panhandle bought their own gas, but Panhandle refused to transport it. [FN209] The Seventh Circuit Court of Appeals characterized the contract between Panhandle and its customers as an exclusive dealing contract that required those customers to purchase all of their natural gas requirements from Panhandle. [FN210] The State of Illinois sued Panhandle both as a natural gas consumer and as parens patriae for other consumers. [FN211] The State pressed both traditional monopolization claims and an essential facilities claim. [FN212]
The court initially noted that essential facilities liability does not exist if competitors can develop competing facilities and if the owner of the essential facility cannot feasibly provide access to that facility. [FN213] The court proceeded to find that access to Panhandle's pipeline was not essential because it would have been economically feasible for competitors to duplicate Panhandle's system by making interconnections between competing pipelines and constructing new pipelines. [FN214] Moreover, Panhandle's refusal to carry competing gas on its pipeline was justified by its exposure to extremely large liability for doing so under its long-term contracts. [FN215]
Essential facilities liability cannot exist unless the owner of the essential facility is a competitor of those entities that the owner excludes. [FN216] In most parts of the information infrastructure, status as an essential facility may be difficult to establish because of the many alternative facilities available. [FN217] The only exception might be a central name server or a major router operating as a joint venture.
b. Essential facilities and public information . Antitrust essential facilities doctrine also might invalidate exclusive dealing arrangements relating to dissemination of public information such as municipal ordinances, state and federal statutes, and judicial and administrative opinions and orders. For these types of information, there may not be alternative sources. This is clearly true if the originator, like a legislature, a court, or an agency, asserts a copyright or enters into an exclusive dealing arrangement. In this circumstance, the competitive market defense is unavailable with respect to the supply of the information in raw form. Any justification for the arrangement would have to arise from its procompetitive effects or the lack of feasibility of using other less restrictive arrangements. Such conditions are unlikely unless the downstream supplier's market is so small that it comprises a natural monopoly. For example, a very small municipality might be able to establish, as a matter of fact, the absence of sufficient demand for its ordinances to induce the participation of more than one supplier of electronic publishing value like chunking and tagging, pointers, and location and distribution. The difficulty with this defense is that the presence of a plaintiff challenging an exclusive arrangement with the purported natural monopolist belies the argument that there is room for only one producer. State action immunity, however, might shield some exclusive arrangements for dissemination of state or local information. [FN218]
7. Application of antitrust doctrines to likely NII ventures
The antitrust doctrines considered in the preceding sections can be summarized and synthesized usefully by applying them to two types of ventures: a jointly owned and operated router and an exclusive vertical arrangement.
a. Jointly owned and operated router . A jointly owned and operated router is subject to antitrust attack on several grounds. First, an agreement among competitors jointly to operate the router is implicitly an agreement for them not to offer their own routers unilaterally. It is thus an agreement to limit supply, such as exists in many joint ventures.
The nature of routers, however, makes it impracticable to have a proliferation of routers which are unilaterally operated. For the architecture *88 to work at all, the degree of cooperation among the unilateral suppliers of routers and routing services would have to be so great as to be equivalent to the cooperation involved in a joint venture router. Thus, the purported supply lost because of the joint arrangement is illusory. Indeed, a jointly operated router may in fact enhance consumer access to all of the participants and thus may increase rather than limit competition among them. Procompetitive justifications figure prominently in the analysis of joint ventures. [FN219] It is unlikely that the joint router would result in liability as an agreement to limit supply.
The router agreement might be attacked as a concerted refusal to deal by anyone excluded from use of the router or disadvantaged by the way that it operates. [FN220] Such an attack could succeed if suppliers of relevant network services could show that they are excluded from participation in the joint venture even though they are willing to pay and meet other terms satisfied by the joint venturers. On the other hand, if the only entities excluded are those refusing to pay the usual price for membership in the venture and otherwise to comply with the terms of the venture, their claims for a concerted refusal to deal almost certainly will fail. The likely factual argument will center on whether the terms offered are equitable, considering the different situations of the various actual and would-be participants. [FN221] Further, an essential facilities claim would fail under application of the same standards discussed with respect to concerted refusal to deal. [FN222] Additionally, neither the concerted refusal to deal nor the essential facilities challenges could succeed unless the challenger could show an absence of competitive alternatives to the jointly operated router. [FN223]
*89 b. Exclusive vertical arrangement . The kinds of vertical arrangements likely to exist in the NII would be unlikely to produce antitrust liability. While the nature of Internet architectures facilitates technological unbundling, [FN224] the market for any particular type of value is likely to be quite competitive. Thus, a producer excluded from access to any particular type of value by a particular exclusive dealing arrangement can find other alternative sources of that type of value. The relevant adjacent markets for infrastructure services resemble the markets for health care services that led to rejection of antitrust claims in that context, except that the infrastructure markets are likely to be even more competitive than the health care markets. [FN225]
The only interesting possibility for liability would be exclusive vertical dealing arrangements relating to public information, like municipal ordinances, state and federal statutes, and judicial and administrative opinions and orders. For these kinds of information, alternative sources may not be available. This is clearly true if the originator, like a legislature, a court, or an agency, asserts copyright or itself enters into an exclusive dealing arrangement.
For example, the Washington State Court System has entered into an exclusive arrangement with an electronic publisher for dissemination of its judicial opinions. Under the arrangement, the court system denies access to the opinions in their original electronic form to competing distributors and publishers. The Securities and Exchange Commission has entered into a similar exclusive arrangement with a dissemination subcontractor, albeit under statutorily mandated conditions that seek to assure the dissemination subcontractor equal access to the product. Nevertheless, the arrangement excludes those who wish to compete with the dissemination subcontractor during the term of the contract, and has been criticized by public interest groups.
In such circumstances, the competitive market defense is unavailable with respect to the supply of the information in raw form. Any justification for the arrangement would have to arise from its procompetitive effects or the lack of feasibility of other less restrictive arrangements. Procompetitive arguments would likely resemble those advanced under essential facilities analysis.
F. Condemnation and Eminent Domain with Respect to Intellectual Property
Denials of access may sometimes follow from the simple assertion of intellectual property rights. [FN226] The holder of a copyright in software or *90 information content, or the holder of the patent in a computer system or process, may deny others the use of the content, software, or system, or may charge prices for its use that the entity desiring access finds unaffordable. The Copyright and Patent Clause of the United States Constitution [FN227] and the copyright and patent statutes acknowledge the legitimacy of such denials of access in the vast majority of cases. [FN228] Indeed, there is no duty to license copyrights or patents under the intellectual property laws or antitrust law. [FN229] Rare circumstances may arise, however, in which legitimate information infrastructure goals cannot be attained without forcing intellectual property owners to grant access. In such circumstances, one source of access rights might be partial condemnation, through the exercise of eminent domain powers by the federal government. [FN230] Indeed, there is a statutory provision in governmental contract law contemplating condemnation. The condemnation approach, while representing significant intrusion into the prerogatives of the intellectual property owner, nevertheless may balance the interests of grantor and grantee of access because it envisions judicially determined compensation for the access.
One important limitation on the condemnation approach is that it may be used only for public purposes, while the interests of the entity granted access through condemnation may largely be private. [FN231] On the other hand, the history of condemnation and eminent domain in the telecommunications industry shows that a mixture of public and private objectives can support the taking of private property, and a sufficient public interest certainly could exist to justify condemning strategically situated intellectual property in the NII context. [FN232] Beyond the public *91 purpose concern, a significant disadvantage to the condemnation approach is the uncertainty and cost of determining compensation through a judicial proceeding. [FN233] However, it could be argued that infrequent use of the condemnation power should make the transaction costs of judicial compensation setting tolerable.
Compulsory licensing is akin to condemnation. Congress has imposed compulsory licensing obligations in the limited case of computer chip mask works. [FN234] Some commentators have proposed compulsory licensing as a useful approach in balancing the interests involved in extending copyright protection to functional features of computer software. [FN235]
The recent controversy over the patents for the RSA encryption algorithm exemplifies one candidate for forcing access through condemnation. Patent holders for the algorithm took the position that certain proposed governmental encryption standards would infringe their patent, and that they should be entitled to royalties for use of the governmentally prescribed standard.
Another possible application for the condemnation doctrine would be in a situation where someone belatedly asserts a copyright interest in basic infrastructure standards like TCP/IP, World Wide Web, or Gopher. In such a circumstance, challenging issues would arise with respect to the nature of the intellectual property, requiring courts to sort out complex problems of intellectual property law before there would be any possibility for condemnation. For example, the RSA patent may be held invalid. Further, the core Internet standards may represent uncopyrightable functions rather than protectible expression. Even assuming the core standards represent protectible expression, they may have been placed in the public domain or subject to broad fair use privileges. [FN236] In situations such as the RSA case, the condemnation approach might be attractive even when a unilaterally determined royalty arrangement is unattractive. Condemnation is a way of forcing an intellectual property owner to grant access when he otherwise would completely refuse access. It is also a way to involve the law in circumscribing pricing for access.
As with other sources of access rights, the appropriateness of condemnation of intellectual property may depend on the availability of alternatives. Alternatives are more likely to be available to copyrighted elements rather than to patented elements, because copyright law allows independent creation. [FN237] In addition, the scope of copyright law for functional *92 works is circumscribed when few or no alternatives exist to the copyrighted work. [FN238] Patent law, on the other hand, forecloses independently created inventions within the scope of the patent. Therefore, it is much more difficult to avoid a patent in seeking to achieve the same function as the patented work.
V. PRIVILEGES AND IMMUNITIES
A. The First Amendment
The First Amendment entitles one to refrain from speaking as well as entitling one to speak. Imposing access duties on a provider of infrastructure services potentially offends this guarantee. [FN239] On this topic, a relevant holding is Turner Broadcasting System, Inc. v. FCC . [FN240] In that case, the United States Supreme Court invalidated a lower court decision upholding cable television "must carry" rules. The Supreme Court remanded to the lower court for development of a better record as to whether the survival of local television stations was jeopardized by unrestricted cable television activity, and whether there were less restrictive means of achieving governmental interests in assuring an outlet for local broadcasting. [FN241] The Turner Broadcasting analysis was based on the First Amendment Privileges and Immunities Clause rather than the Fifth Amendment Takings or Due Process Clauses. [FN242]
The First Amendment can also support claims of access rights. Usually, refusal of access by a private network services provider cannot support a claim for violation of the constitutional rights of the entity desiring access because the refusal is not "state action." In Altmann v. Television Signal Corp. , [FN243] however, a federal district court found that sections 10(a) and (c) of the 1992 Cable Act sufficiently encouraged private cable operators to ban constitutionally protected material from leased and public access channels to justify an injunction against an outright ban. [FN244] The district court made this finding while acknowledging the general rule that state regulation of a private industry usually is not enough to establish *93 state action. [FN245] T