Making the World Wide Web Safe for Democracy:

A Medium-Specific First Amendment Analysis

Andrew Chin [FNa1]

Reprinted from 19 Hastings Communications & Entertainment Law Journal 309 (1997).   Permission is hereby granted to download and/or print one copy for personal use.   Republication in any form, including electronic, without written consent of the author is prohibited.

© 1997 Hastings College of the Law; Andrew Chin.

Introduction

Last summer's district court decision in ACLU v. Reno, [FN1] enjoining enforcement of the Communications Decency Act of 1996, [FN2] was hailed by the on-line civil liberties community as "a resounding victory for First Amendment rights everywhere." [FN3] The lead attorney for the plaintiffs praised the ruling as reflecting "an enormously sophisticated understanding of the Internet," and predicted that it would "guide all future courts in deciding a whole variety of issues that apply to the net." [FN4] A closer examination, however, reveals the court's "understanding" to be a simplistic view of the distributional consequences of the Internet that foreshortens the First Amendment analysis and constrains the Internet's prospects for fulfilling its democratic promise.

Judge Stewart Dalzell's opinion, which includes what he labels a "medium- specific" First Amendment analysis, [FN5] is illustrative. Characterizing the Internet as a "never-ending worldwide conversation" and "the most participatory form of mass speech yet developed," Dalzell concluded that the global computer communications network "deserves the highest protection from governmental intrusion." [FN6]

The purpose of this Article is to suggest that the Internet's unprecedented communicative potential as a cheap, vast, interactive forum does not, by itself, render governmental regulation of its content constitutionally suspect. In particular, the structural impact of the World Wide Web (Web) on the distribution of power in public discourse may justify intervention by the state. To the extent that the Web's free market of homepages and links amplifies the voices of the powerful and silences the powerless, impoverishing public debate, corrective policy measures should be constitutionally favored.

This Article presents a medium-specific analysis, both theoretical and empirical, of the structural impact of the Web on public discourse. Section I provides an introduction to the Web and the related terminology used throughout this Article. Section II surveys the doctrinal context for the analysis in this Article. Section III uses a mathematical model of the Web's link structure relative to groups of speakers to identify the significant factors affecting the distribution of communicative power on the Web. Section IV surveys the emerging structure of public discourse on the Web. Section V proposes egalitarian policy reforms based on the analysis in this Article, and assesses their constitutionality. Section VI reviews ACLU v. Reno from the perspectives elaborated in this Article.

I. The World Wide Web

The Web is a collection of more than 30 million documents stored in different computers throughout the Internet. These documents may incorporate any combination of text, graphics, audio, video, computer programs, and/or any other data that can be stored on digital media. A user of a networked computer can publish a document on the Web simply by copying it into a specially designated, publicly accessible directory on the computer. The document is then immediately available to anyone on the Internet who knows the address of the computer and the name of the document.

Thus, an individual or group can establish a presence, or "site," on the Web by publishing a set of documents representing the publisher's viewpoint or providing other information relating to the publisher. Typically, one of the documents is the "homepage," or the first access point to the site. The homepage usually provides an overview of the site and references, or "links," to the other documents on the site. Additionally, a Web document may also include any number of links to documents stored anywhere on the Web. Thus, links can be used as cross-references within a single document, between documents on the same site, or between documents on different sites.

By using a special computer program called a "Web browser," a reader can view, or "browse," any Web document. The reader can specify a document to browse by typing in its address, selecting it from a personal "bookmark file" of frequently used addresses, or following a link from the document currently being viewed to the document being referenced. When a document is accessed by a reader, it receives a "hit." For example, within the span of a few seconds, a reader may type in the Web address of a political party to retrieve its homepage, follow one of the links on that homepage to a document listing some of the political party's allies, and then follow one of those links to the homepage of an allied organization. Each of these three documents would be credited with one hit.

A significant consequence of the Web's structure, then, is that publishers can influence a reader's access to other speakers on the Internet through their selective inclusion of external links. While it is true that other forms of speech may include references to other speakers, the Web is unique in that references, even to speakers halfway around the world, can be accessed as easily as cross-references within works by the same author. In a practical sense, linked documents are truly incorporated by reference as if presented by the same author.

The set of external links included within a document is part of the expressive content of that document. The external links also substantially influence the content received by the reader in subsequent document accesses. Generally, readers explore, or "surf," the Web by following links. Influenced by external links, this process often leads users to discover subject matter and perspectives unrelated to their original browsing objectives. [FN7]

For these reasons, a medium-specific analysis of speech on the Web must examine the distribution of the links among its sites and speakers. As the next section will suggest, the structure of links on the Web, while generated through free self-expression, may concentrate speech power in a manner that contradicts the democratic principles of the First Amendment.

II. The Doctrinal Context

The Web is only one of many fora where social structure constrains public discourse, and for which state intervention has been urged. Madisonian First Amendment scholarship, informed by values of liberty and by principles of equality, has repeatedly recognized the inability of the free market to satisfy the nation's constitutional commitment to robust public debate. The works of Alexander Meiklejohn, [FN8] Cass Sunstein, [FN9] Owen Fiss, [FN10] Stephen Gardbaum, [FN11] Frederick Schauer, [FN12] and Morton Horwitz [FN13] offer coherent First Amendment analyses recognizing the central importance of speech furthering self-governance. [FN14] In this analysis, the state can and should regulate public discourse to the extent that the free market constrains its quality. As Fiss describes, "The state is to act as the much-needed countervailing power, to counteract the skew of public debate attributable to the market and thus preserve the essential conditions of a democracy." [FN15]

The Madisonian perspective has made a few appearances in the rapidly expanding literature on First Amendment issues in cyberspace, but, thus far, it has only sounded tentative cautions against the excesses of consumer choice. Sunstein writes:

Fiss echoes these hopes and fears:

The Madisonian perspective has not yet fully addressed the Web's emergence as the leading mode of mass communication in cyberspace. In the meantime, however, the courts have been leaving no room on the Internet for the Madisonian view of the First Amendment. In the past, the Supreme Court allowed content regulation only of electronic media that presented structural restrictions on access (Red Lion [FN18] and Pacifica [FN19]) or transmission (Turner Broadcasting [FN20]). Already though, several First Amendment cases concerning the Internet have distinguished Red Lion, Pacifica, and Turner Broadcasting by observing that these scarcity rationales no longer apply. [FN21]

To make room for Madison in cyberspace--to make the Web safe for democracy-- requires a new, particularized rationale for content regulation. If democratic deliberation is to be a First Amendment value in cyberspace, concerns about balkanization and concentration of power must be expressed in medium- specific terms, and preferably based on structural as well as empirical analysis. As the Supreme Court has held, "[e]ach medium of expression ... must be assessed for First Amendment purposes by standards suited to it, for each may present its own problems." [FN22] Although the preliminary writings of Sunstein and Fiss do not provide a basis for egalitarian content regulation on the Web, they raise appropriate questions for a medium-specific First Amendment analysis. Does the Web promote or hinder democratic deliberation? Does the Web balkanize the social order in cyberspace? Do Web sites engage publishers and readers of opposing viewpoints, or do they merely provide "what we already agree with?"

A medium-specific analysis would begin by examining the state of the Web with respect to the structure of public discourse. For illustration, suppose that there are two perspectives, A and B, with respect to a particular political issue. Perspective A is held by 40% of the public and Perspective B is held by 10%, with the remaining 50% undecided. Each perspective is represented by a number of sites on the Web, proportional to its level of support in the population. Suppose that proponents of B believe that their perspective will be persuasive to anyone who engages in a deliberative comparison between A and B. Web sites for B therefore include many links to Web sites for A. On the other hand, proponents of A may believe that the best way to protect their lead in the polls is to avoid any reference to B. Because there are many more A sites than B sites on the Web, publishers of A sites can be confident that their perspective will be seen by the undecided reader.

As a result of these strategies, Web sites for A actually garner more than four times as many hits as Web sites for B among exploring readers. Web sites for A benefit from their opponents' referrals without returning the favor. As a result, the Web serves to concentrate speech power in the hands of the majority. This situation is analogous to the plight of marginalized groups in conventional public discourse: the minority group, in order to survive, must understand the dominant perspective sufficiently to deconstruct and criticize it, whereas the mainstream group may benefit unjustly from its ignorance of minority perspectives.

To the extent that the above scenario is descriptive of the political landscape in cyberspace, the concerns of Madisonian First Amendment scholars weigh heavily against Judge Dalzell's sanguine characterization of the Web's effects on free speech. Drawing on the postulated terms, the remainder of this Article provides a sobering reassessment of the Web's effects on the structure of public discourse.

III. A First Amendment Analysis of the Web

If the courts are to assess the validity of state intervention by asking "whether the intervention in fact enriches rather than impoverishes public debate," [FN23] the analysis must be made tractable in the context defined by the medium. Part of the analytical difficulty of this structuralist approach to the First Amendment, Fiss notes, is that "[w]e must be certain everything worth saying is said" without prejudging "what is worth saying." [FN24] To be sure, "there is no such thing as a false idea"; [FN25] nonetheless, any equality principle that might justify content regulation must face the empirical reality that some ideas simply do not deserve equal time. [FN26] It is possible, however, to proceed with the analysis without having globally identified all ideas worthy of a hearing or all Web documents worthy of hits. By "proceed[ing] in a negative fashion, trying to identify impermissible effects ('group disadvantaging,' 'disproportionate impact')," judges can recognize and analyze forces that distort public discourse, even "without a commitment to a particular end-state." [FN27]

In that spirit, I propose a quantitative model that represents a reviewing court's inquiry as to whether a given actor enriches or impoverishes public debate. This model captures the Madisonian concern with the concentration of speech power without identifying a priori entitlements to space in the public square. This model is applicable to any medium with quantitative measures of speech power. Furthermore, I believe it supplies an analytical construct necessary to weigh the libertarian impulses of Dalzell against the egalitarian impulses of Sunstein and Fiss with respect to the Web. Specifically, the model identifies the analytical factors relevant to deciding whether the Web is a structural impediment to robust public debate.

A. A Quantitative Model of Rich Public Debate

The proposed model characterizes the public square as a speech domain comprising several sets of speakers, defined so that each set possesses an equal entitlement to participate in public debate. Given a quantitative measure of speech power, we will employ a concentration index that is based on the standard deviation as the measure of inequality across the sets. [FN28] An actor that increases the inequality of the distribution of speech power will be said to impoverish public debate; an actor that decreases inequality thereby is said to enrich public debate.

Note that, in the foregoing model, we have not assigned any semantic or normative interpretations to the sets of speakers. We do not, for example, assume that Republicans and Democrats deserve equal time. Nor do we suppose that a group comprising a given percentage of the populace is entitled to a proportionate share of the public square; to do so would presume that the group had something worthwhile to say. The sets are likely to be heterogeneous and will vary in size. Most significantly, we do not assume any relationship between the sets and any specific content or viewpoint, so that regulations based on the structural analysis of inequality will not be content-based. Indeed, the sets of speakers may not correspond naturally to any recognizable political or cultural partition of society. [FN29] The point of the model is to quantify changes in the concentration of speech power across all of society in the context of a particular medium, not to support the claims of any particular subgroup.

The standard deviation provides a measure of inequality that takes into account the entire distribution of speech power. Fiss's equality principle, that the least powerful speakers should receive the highest level of First Amendment protection, [FN30] also aims at minimizing inequality, but may fail to recognize both opportunities for, and threats to, robust public debate. For example, the distribution (50, 20, 19, 11) is markedly more unequal--and has a higher standard deviation--than the distribution (30, 30, 30, 10), but it is not clear from Fiss's rule that the latter distribution is preferable.

B. A Mathematical Model of the Web

Having formalized a model for measuring the inequality of speech power across a society, we can analyze the impact of the Web on those inequalities.

[Mathematical derivations omitted.  See original in Hastings Comm/Ent for the equations, or check back here for an update in a few months.]

Thus, the equality principle we have proposed would favor granting the highest level of First Amendment protection to those Web sites that (1) provide an equitable distribution of links to all other sets of speakers, in accordance with their equal speech entitlements; and (2) rely upon (i.e., attract links and hits from) an unorthodox cross-section of Web publishers and readers. Such sites contribute to robust public debate by presenting a fair number of links to other sets of speakers, but are the most vulnerable sites on the Web because they must rely upon a base of support that is contrary to prevailing market preferences.

Returning to the example given in Section II, we find that the B sites would be favored under the First Amendment. By publishing links to their opponents' sites, the B sites seek to provide an equitable distribution of links. Also, because B sites do not receive any links from A sites, they must rely upon an unorthodox cross-section of Web publishers and readers. On the other hand, A sites would not be favored. By refusing to include links to their opponents, the A sites do not provide an equitable distribution of links. Further, because A sites receive links from a proportionate number of B sites as well as A sites, they benefit from a representative cross-section of Web publishers and readers.

An obvious remedy for this inequity is to require at least some of the A sites to provide links to B sites. We will discuss this "must-carry" proposal in detail in Section V. First, however, we will establish the case for reform by analyzing the current state of the Web with respect to the two criteria we have developed in this section.

IV. The State of the Web

Since its introduction to the public in 1991, the Web has accompanied the near-exponential growth of the Internet to become a true mass medium. Today, Web site addresses are commonly used in advertising, promotions, and corporate identification, and some 600,000 individuals have published personal homepages. [FN31] An estimated 11.5 to 18.2 million Americans use the Web, [FN32] which already accounts for more traffic on the Internet than any other network application. [FN33] A leading software industry executive speculates that within the next five years, there will be at least 500 million devices capable of browsing the Web. [FN34]

The economies of the Internet, along with the flexible structure of Web documents make the Web the most likely trajectory for the long-awaited "convergence of communications and media." [FN35] Already, MSNBC, CNET, CBS, CNN, and various radio stations have demonstrated the robustness of the Web model by "webcasting," or making some of their transmissions available as real- time audio and/or video performances within published Web documents. [FN36] Many newspapers and magazines also publish electronically via the Web. [FN37] Web browsers have already incorporated electronic mail facilities into their services, and will soon allow real-time voice communications as well. [FN38] Televisions and telephones capable of browsing the Web will be available to the public this year. [FN39] A scenario in which the Web swallows all other communications media no longer seems merely plausible, but probable.

For the foreseeable future, the state of the Web will be subject to fundamental, continual, and far-reaching change. Every second of the day, Web documents are published, edited, moved, and deleted, all without authorization by any governing body or notification to any central registry. Thus, any empirical description of the Web is literally obsolete as soon as it is written, and any predictions based on the current state of the Web must rely heavily on conjecture.

This change and uncertainty does not preclude an investigation of the Web's present-day features that are relevant to free speech. By analyzing the state of the Web with respect to First Amendment criteria, we can assess the extent to which the Web is already exhibiting anti-democratic properties, identify areas of particular concern, and suggest corrective action where necessary.

A. Corporate Domination of the Web

As we have already described, two indicators of speech power on the Web are hits and links. Each hit represents an instance of communication in which a document is accessed. Thus, the effective speech power of a Web site may be roughly measured by the aggregate number of hits received by its documents. Speech power is also affected by the distribution of links. With other factors equal, a site that receives many referrals from other sites is more likely to be visited than a site that receives fewer referrals.

Data on hits and links is collected and published on the Web, but is not completely reliable. Hits are especially difficult to measure because many sites do not keep track of accesses, the method of counting accesses may vary from site to site, and sites may inflate their access statistics. Link counts are easier to validate, but are less closely associated with speech power than hit counts. Despite this imprecision, it is possible to draw two general conclusions from the available data. First, speech power on the Web is already largely dominated by corporations, primarily computer companies and broadcasters. Second, to the extent that speech power is available to individuals, it is dominated by pornography and content derived from corporate providers, rather than self- expression and public discourse.

Presented in Tables 1 and 2 are examples of reliable "Top 25 Sites" lists based respectively on the hits and links criteria. Table 1 lists the 25 sites that received the most hits during July 1996, as reported by the "100hot" Web site. [FN40] Table 2 lists the 25 sites that were most frequently linked to by other sites on the Web as of April 14, 1996, as reported by the "WebCrawler" site. [FN41]

Table 1. Most-Visited World Wide Web Sites, July 1996 

Table 2. Most-Linked-To World Wide Web Sites, April 1996 

The "100hot" Web site also provides rankings of the most-accessed personal home pages on and off college campuses (as identified by the ".edu" Internet domain). [FN42] Of the top ten college sites in July 1996, six were image libraries of celebrities and fashion models derived from print and broadcast media. [FN43] All ten of the top off-campus sites were sexually explicit image libraries. [FN44]

Two terms of art appearing in Tables 1 and 2, "Web directories" and "Web search engines," require elaboration. Directories and search engines are sites that provide referral services for Web users. Web directories classify other Web sites by subject matter and provide other summary information relating to each site. Web search engines allow users to locate textual occurrences of terms within a database of Web documents. These services allow users to find desired information on the Web without knowing a specific Web address in advance. They are typically free and easy to use, and consequently very popular.

Both directories and search engines gather their data by using special Web browsing software that traverses every link found on the Web. This process is so computationally intensive and imposes such massive storage requirements that general-purpose Web directories and search engines have been provided only by the for-profit sector. It is foreseeable that these costs will be recovered. For example, Alta Vista, [FN45] a leading Web search engine owned by Digital Equipment Corporation, will ultimately be bundled with browser and other communications software as a transparent, value-added component. As the controller of one of the most convenient entry points to the Web, Digital hopes to create and benefit from an industry standard for searching, in the same way that Netscape defined the standard for more general browsing. [FN46]

Notably, comprehensive Web search services effectively provide links to every document on the Web, thereby satisfying at least the first of the egalitarian principles outlined in the previous section. In fact, search engines are an essential component of the policies proposed in Section V. In view of commercial developments, however, it is premature to assume that existing Web search services will not ultimately result in increased corporate control of speech power.

B. Balkanization of Political Discourse

Political content occupies a relatively small but vital segment of the Web. In 1996, political speech on the Web exhibited a rich diversity of perspectives and formats. Political speakers, however, tended to discourage public debate by refusing to provide links to opposing viewpoints. [FN47] I performed a content analysis of 116 political Web sites [FN48] and the links to and from these sites. The sites were taken from the "Political Site of the Day Archives" [FN49] for the six-month period from August 1995 to February 1996, and were visited during the week of July 15-19, 1996. [FN50] The results of this analysis are summarized in Tables 3 and 4.

Table 3. Prevalence of Links Between Political Web Sites, By Political Orientation, July 1996  

This data presents a bleak vision of democratic discourse on the Web. As Table 3 shows, only a small minority of partisan sites (12 of 79, or 15%) offered links to opposing viewpoints. Moreover, even in this small sample, the majority-minority power dynamics described throughout this Article are noticeable. Taken in aggregate, the sites offering mainstream perspectives -- Democrats, Republicans, and Libertarians -- provided less diversity than they received in referrals. On the other hand, the Progressive and "Other" Web sites offered more links to opposing views than they received in return. Table 4 shows that these patterns were also common to various categories based on site content.

Table 4. Prevalence of Links Between Political Web Sites, By Content Type, July 1996    

These observations challenge the premise that the Web's contribution to free speech can be measured solely in terms of the number and diversity of its speakers. Far from fostering deliberative political discourse, most of the surveyed Web sites sought to consolidate speech power and served to balkanize the public forum.

C. Concentration of Media Power

The corporate domination of the Web and the balkanization of social discourse have arisen in an era of increasing concentration in the media and telecommunications industries. [FN53] Some business analysts predict that mergers and acquisitions in these fields will continue until as few as six to ten global conglomerates control these markets. [FN54]

The Telecommunications Act of 1996 [FN55] cleared the way for much broader cross-ownership of local and long-distance telephone services, and broadcast and cable television. While this deregulation was intended to encourage the efficient convergence of communication technologies, [FN56] it is also likely to accelerate the concentration of media power.

As the prevalence of personal Web sites devoted to celebrities and fashion models [FN57] illustrates, corporate domination of the media extends itself into the Web. As Robert McChesney points out, it is likely that the undemocratic hierarchies of speech power in the traditional media will ultimately be reproduced in cyberspace:

V. Free Speech Policies for the Web

I have identified four policy initiatives that would significantly improve the prospects for democratic discourse on the Web. They are described below, listed in increasing order of ambition and scope.

A. Free Public Search Engines and Directories

As a first step toward ensuring a fair distribution of links to all Web sites, the federal government should provide comprehensive search engines and directories that are free to the public. These services would benefit Web publishers by providing starting points on the Web from which all other sites would be equally accessible. The search engines and directories would also allow users an opportunity to access and explore the Web without being inundated by corporate advertisements and promotions.

Free public search services are consistent with longstanding government policies favoring universal access to information. Just as postal subsidies have made newspapers more available and free public libraries have made books more available, the government has an important role in making Web pages more available.

B. Free Public Link Exchanges

The federal government should also provide a service that allows Web publishers to exchange links with each other on a site-for-site basis. To participate, a publisher would reserve a space on her home page for a "guest" link, and then inform the link exchange that she had taken this action. In return, her page would be advertised on other sites, which would change at specified time intervals. Because all participating sites would appear in the same rotation, the link exchange would have the democratically favored effect of providing the most references to the sites receiving the fewest hits. Even so, all participating sites would benefit to some degree from the exchange.

The Internet Link Exchange ("ILE"), [FN59] a for-profit service, highlights the distinctive features of the proposed exchange and the need for governmental intervention. Links on the ILE are exchanged on a "two hits-for- one hit" basis, so that the least-read sites derive the least benefit from the program. The ILE takes half of the hits as overhead, using them for commercial advertising. [FN60] The ILE's structure thus reflects and amplifies the anti- democratic rationality of the marketplace.

C. Must-Carry Regulations

To counteract the structural forces tending to concentrate speech power on the Web, [FN61] Congress should take the further step of establishing must- carry rules governing the most popular sites. For example, sites receiving more than 1,000 hits a week could be required to carry at least five links to sites participating in the free public link exchange.

The exact parameters of such a policy will require a more comprehensive and precise review of the state of the Web than has been presented here. In any case, however, the statistics in Section IV and the structural analysis in Section III demonstrate the need for substantial redistribution of speech power if the Web is to fulfill its democratic promise. As Edwin Baker suggests, "[t]he aim [in media regulation] should be structure that promotes creative opportunities and facilitates audience access to diverse cultural, partisan, and informational communication." [FN62]

A must-carry rule for the Web, [FN63] based solely on the level of traffic at a site and not on any attributes of its publisher or its content, is a content- neutral regulation. [FN64] This regulation compares favorably with the Federal Communications Commission rules that were analyzed in the Supreme Court's first review of the Turner Broadcasting [FN65] case. It should also be found constitutional under the O'Brien [FN66] test for content-neutral speech regulation. In O'Brien, the Supreme Court held that a content-neutral regulation passes First Amendment scrutiny if "it furthers an important or substantial governmental interest; if the governmental interest is unrelated to the suppression of free expression; and if the incidental restriction on alleged First Amendment freedoms is no greater than is essential to the furtherance of that interest." [FN67]

In Turner Broadcasting, the Court held that "assuring the public has access to a multiplicity of information sources is a governmental purpose of the highest order, for it promotes values central to the First Amendment." [FN68] Accordingly, provisions of the Cable Act of 1992 requiring cable television systems to reserve a fraction of their channels for local broadcast television stations would be found to serve an important governmental interest if the government could show that "1) unless cable operators are compelled to carry broadcast stations, significant numbers of broadcast stations will be refused carriage on cable systems; and 2) that the broadcast stations denied carriage will either deteriorate to a substantial degree or fail altogether." [FN69]

The proposed must-carry rule for the Web would impose an analogous regulation on sites having significant power in a concentrated speech market. Our analysis strongly suggests that this regulation, as applied to the Web, would satisfy both of the required showings. First, our survey of political Web sites in Section IV indicated that significant numbers of less powerful sites are denied links from more powerful sites. Second, our structural analysis in Section III demonstrated that the absence of such links further concentrates speech power on the Web, so that the less powerful sites measurably "deteriorate to a substantial degree."

With respect to narrow tailoring, the Court in Turner Broadcasting held that "a regulation need not be the least speech-restrictive means of advancing the government's interest." [FN70] Rather, the regulation should not "burden substantially more speech than is necessary to further the government's legitimate interests." [FN71] Thus, on remand, the FCC regulations would be found to be narrowly tailored if the burdens on cable operators and cable programmers were not substantially greater than necessary. Such burdens would include:

None of these burdens is seriously at issue under the proposed Web regulations. Web documents can be any size, subject only to the constraints of disk space on the host computer. Thus, adding links to a document does not displace any of its content. A Web document can easily be augmented with five links using less than one kilobyte of disk space. It is inconceivable that a site attracting more than 1,000 hits a week would be located on a host machine that did not have sufficient available disk space to accommodate the must-carry links.

Similarly, the proposed Web regulations would also be unrelated to the suppression of free expression. A must-carry rule would augment speech, not displace it. Like the FCC provisions, the Web regulations would be "broad- based" and, therefore, would not be "structured in a manner that carries the inherent risk of undermining First Amendment interests." [FN73] Therefore, it is safe to conclude that a must-carry rule for the Web would survive First Amendment scrutiny under Turner Broadcasting and O'Brien.

The Court in Turner Broadcasting did not focus solely, or even primarily, on the number of existing broadcast television stations. Instead, it recognized that the market power of cable television systems created medium-specific issues that required a more sophisticated First Amendment analysis. [FN74] Similarly, courts should recognize that the structure of the Web presents special First Amendment problems that are not properly addressed solely by assertions about the ease, popularity, and diversity of Web publishing. As this Article has demonstrated and as Eugene Volokh has observed, "a greater diversity of available speech need not lead to the diversification of what is actually consumed." [FN75]

D. Content-Based Regulation

The focus on content neutrality in First Amendment jurisprudence presents a formidable obstacle for content-based regulation of any speech medium. To survive strict scrutiny, any content restriction in a public forum must be narrowly tailored to serve a compelling government interest. [FN76] For example, prohibitions against pornography have been found invalid for overbreadth to the extent that the definition of "pornography" differs from that of "obscenity" and includes protected speech. [FN77]

An alternative First Amendment doctrine, based on Madisonian principles, would recognize that "[t]he state might ... have the right to stop the general advocacy of an idea when that advocacy has the effect of interfering with the speech rights of others." [FN78] In applying this doctrine to a restriction on pornography, a court would weigh the silencing effect of pornography on women [FN79] against the speech value of the prohibited materials. Such an analysis would probably support broader regulations on speech protection than what would withstand strict scrutiny under current law. In particular, given the prevalence of pornography on the Web [FN80] and the vast gender disparity among Web authors, [FN81] the silencing effect of pornography on women may be more easily documented on the Web than in traditional fora, thus justifying specific governmental intervention.

VI. ACLU v. Reno Revisited

Even without supposing a Madisonian transformation of First Amendment doctrine as described above, [FN82] it is possible to criticize ACLU v. Reno on its own terms. Under current free speech doctrine, the court correctly ruled that the challenged provisions of the Communications Decency Act of 1996 (CDA) [FN83] were unconstitutionally overbroad. [FN84] Despite the best efforts of Judge Dalzell, however, the district court failed to provide a medium-specific First Amendment analysis of Internet regulation. [FN85]

The CDA was signed into law on February 8, 1996. Comprising Title V of the Telecommunications Act, the CDA contains provisions making it a crime to make "indecent" or "patently offensive" materials available to minors over the Internet. [FN86] On the same day, the ACLU and 20 other organizations filed actions in the Eastern District of Pennsylvania for a temporary restraining order to enjoin enforcement of two provisions of the CDA. [FN87] Soon thereafter, the American Library Association and 26 other organizations filed a similar action. [FN88] The actions were consolidated and heard by a three-judge court consisting of Chief Judge Dolores Sloviter of the United States Court of Appeals for the Third Circuit, and Judges of the Eastern District of Pennsylvania, Ronald Buckwalter and Stewart Dalzell. [FN89] After extensive stipulations of fact, hearings, and oral arguments, [FN90] the court granted the preliminary injunction. The court unanimously found both of the challenged provisions facially unconstitutional, [FN91] but issued three separate opinions with differing analyses.

Judge Sloviter held that although "there is certainly a compelling government interest to shield a substantial number of minors from some" sexually explicit online materials, [FN92] the CDA is overbroad to the extent that it covers some protected speech. [FN93] Because the CDA is content-based, Sloviter noted that it is presumptively invalid unless it is narrowly tailored to serve a compelling government interest. [FN94] From the findings of fact, she found that "it is either technologically impossible or economically prohibitive for many of the plaintiffs to comply with the CDA without seriously impeding their posting of online material which adults have a constitutional right to access." [FN95]

Judge Buckwalter's analysis focused on Fifth Amendment due process considerations. He emphasized that "[i]t is a basic principle of due process that an enactment is void for vagueness if its prohibitions are not clearly defined." [FN96] Noting that the word "indecent" in the CDA "is an undefined word which, standing alone, offers no guidelines whatsoever as to its parameters," [FN97] Buckwalter concluded that the provisions were unconstitutionally vague. [FN98]

Judge Dalzell's opinion, the longest and most eloquent of the three and the only one claiming to take a "medium-specific approach," [FN99] relied heavily on the "very low barriers to entry" for communication on the Internet. [FN100] "[A]s a result of these low barriers," he wrote, "astoundingly diverse content is available on the Internet .... [T]he Internet provides significant access to all who wish to speak in the medium, and even creates a relative parity among speakers." [FN101] Dalzell went on to argue that the costs of compliance with the CDA, [FN102] by raising the barriers to entry, would diminish the diversity of content and "skew the relative parity among speakers ... on the Internet." [FN103] As "the most participatory marketplace of mass speech that this country--and indeed the world--has yet seen," Dalzell concluded, "the Internet deserves the broadest possible protection from government-imposed, content-based regulation." [FN104] Applying strict scrutiny, he accepted the government's interest in protecting children from pornography, [FN105] but found the CDA overbroad to the extent that it covered protected speech. [FN106]

Of the three opinions, Dalzell's opinion is the most troubling from the Madisonian perspective. Sloviter's and Buckwalter's opinions simply apply the established First Amendment overbreadth doctrine to a content-based speech regulation, as they would in any medium. While their analyses do not take into account the interests of those silenced by indecent speech, their failure to do so is the necessary consequence of a jurisprudence centered on content neutrality.

On the other hand, Dalzell attempts to fix new standards for content regulation of the Internet. In conferring "the broadest possible protection" upon the new medium, he prejudges the distributional consequences of the Internet [FN107] and thus minimizes the prospects for egalitarian reform. In particular, Dalzell would probably reject the must-carry rule proposed in Section V.C under strict strutiny. [FN108] This "medium-specific" approach may make Dalzell's opinion a favorite among commentators, [FN109] but it is also a serious threat to the Madisonian aspirations of the Web.

VII. Conclusion

The Web is a vast speech domain that may ultimately swallow all telecommunications media currently in use. Because of its potential scope, the Web presents important opportunities for First Amendment jurisprudence. For example, the convergence of media technologies will enable the courts to discard the distinct "broadcast model" of free speech law and return to more well-established First Amendment principles. [FN110]

An even more exciting prospect is that the Web, because of its vastness, may accommodate the frequently incompatible interests of liberty and equality. The must-carry rule proposed in this Article is an example of a policy that would enhance public debate and expand the diversity of public discourse without sacrificing individual expressive rights. However, given the shape of things to come, it is still much too early for the government to abandon its role in making the Web safe for democracy.

Footnotes