The Hyperlinked Society: Questioning Connections in the Digital Age

MATTHEW HINDMAN

What Is the Online Public Sphere Good For?

Almost from the moment the Internet became a mass medium, observers predicted that it would change the relationship between citizens and the political information they consume. According to numerous accounts, the Internet would function as a digital printing press, enabling any motivated citizen to publish his or her views for a potential audience of millions. The architecture of the Web would also instantly link citizens with diverse opinions to one another. This citizen-created hyperlinked content would not need to follow the biases, whims, and market demands that constrain traditional media. Without barriers to entry, the public sphere would become vastly broader and more representative.

Recent events have borne out at least some of this breathless, mid-1990s Internet boosterism. Internet sources are now a large and still rapidly growing portion of American's diet of political media. According to one recent study, 14 percent of the public relied primarily on Internet sources in the lead up to the 2006 midterm elections, which is double the percentage of four years earlier.¹ Citizens have also rushed to their digital printing presses with an eagerness matching the most optimistic predictions. Surveys have suggested that twelve million Americans maintain a blog, with about 10 percent of these blogs focused primarily on politics.² The most popular political blogs and political Web sites now claim far more readers than traditional opinion journals, such as the Nation, National Review, or the New Republic. In several prominent incidents, stories first reported in political blogs became the focus of sustained mainstream press coverage.

Against this backdrop, some scholars have revised, qualified, and extended early theories about how the Internet would transform the public sphere. Cass Sunstein's recent book Infotopia focuses on new collaborative models that allow citizens themselves to create and filter high-quality political information.³ Richard Rogers similarly suggests that, despite its limitations, the Web is “the finest candidate there is for unsettling informational politics,”⁴ offering citizens exposure to political points of view not heard in traditional media. Many scholars have focused on blogging as reason for optimism. Despite a critical assessment of online deliberative forums, Andrew Chadwick concludes, “The explosion of blogging has democratized access to the tools and techniques required to make a political difference through content creation.”⁵ While Daniel Drezner and Henry Farrell note that some blogs garner far more readership than others, they state, “Ultimately, the greatest advantage of the blogosphere is its accessibility.”⁶

Yet perhaps the most prominent recent account in this vein is Yochai Benkler's The Wealth of Networks.⁷ Compared to traditional media, Benkler suggests, the Internet allows for a broader, more inclusive and more densely linked public sphere. Like the other scholars already cited, Benkler argues that the Internet does not just place far more information in the hands of interested citizens; it transforms public debates by enabling online communities to use collaborative methods to create content, correct inaccuracies, and send readers to the most insightful commentators.

In this essay, I focus on Benkler's influential book to make two central claims. First, I suggest that his vision of the “networked public sphere” is partly correct and that the Internet is strengthening some democratic values. Benkler's account in particular illuminates important aspects of the Internet's impact on collective action, including the way it has made it easier to aggregate small contributions into a useful whole. There is also evidence that the Internet is strengthening public oversight by making “fire alarm” or “burglar alarm” models of citizenship more effective.⁸ Finally, there is reason to believe that the Internet has made journalists and other political elites more accountable or at least more vulnerable. Former Senate majority leader Trent Lott and former New York Times reporter Judith Miller have both publicly laid the blame for their travails on bloggers' criticism.⁹ The Mark Foley scandal that dogged Republicans during the 2006 midterm election seems to have been touched off by an obscure political blog that posted “overly friendly” e-mails between Foley and a former congressional page.¹⁰ In these incidents, the evidence for the Internet's role is strong, and the political consequences have been dramatic.

Second, however, I argue that while the Internet is strengthening some democratic values, it has placed others at risk. Many continue to celebrate the Internet for its inclusiveness; others decry the medium for the same reason, worrying (as CNN president Jonathan Klein put it) that the Internet gives too much power to “a guy sitting on his couch in his pajamas.”¹¹ I argue here that the underlying premise of both assessments is wrong. Inclusiveness is precisely what the online public sphere lacks. Part of the problem is the extraordinary concentration of links and patterns in online traffic. For example, several observers have suggested that a group of A-list political bloggers attract disproportionate attention. I argue here that even the emergence of a blogging A-list barely scratches the surface of online inequality.

I am going to develop this argument by targeting what I term the “trickle-up theory” of online discourse, particularly as it is formulated in Benkler's account of what he terms the “networked public sphere.” As I have already suggested, Benkler's work merits special attention for several reasons. The Wealth of Networks is important in its own right, the culmination of nearly a decade of scholarship; yet Benkler's claims are also representative of those made by others. One of the virtues of Benkler's book is that it fully explicates key claims and assumptions that other scholars often gloss over. Benkler is also scrupulous about cataloging and responding to potential counterarguments.

Like many other observers, Benkler argues that the networked structure of the Web itself can compensate for inequalities in traffic and in the elite profile of those who publish the most read online political outlets. He describes blogs as an “ecosystem,” in which even the smallest outlets have an important role to play. Insights or discoveries made by lower-ranking blogs can (in theory) travel up the hierarchy of online outlets, with the most worthy posts receiving a torrent of attention if they are linked to by the most prominent blogs. His account persistently reframes inequalities in egalitarian terms, recasting them as “collaborative filtering” or “meritocracy” in action. In what follows, I suggest that there are several reasons to be suspicious of the trickle-up theory of public debate advanced by Benkler and others. Thus far at least, public discourse online looks more like a multilevel marketing scheme than a Habermasian ideal.

What Is the Public Sphere For?

Before critiquing recent accounts of the online public sphere, it is worth placing such scholarship in a broader context. In recent decades and well before the rise of the Internet as a mass medium, there has been a resurgence of interest among scholars in the public sphere. Much of the initial credit for this belongs to Jurgen Habermas.¹² Yet what John Dryzek calls the “deliberative turn” in political philosophy now includes numerous theorists: John Rawls,¹³ Joshua Cohen,¹⁴ Carlos Nino,¹⁵ Amy Gutmann and Dennis Thompson,¹⁶ John Dryzek himself,¹⁷ and Bruce Ackerman and James Fishkin,¹⁸ to name a few.

Despite long-running academic disputes, the visions of deliberation authored by this group show strong commonalities. All look to enrich democracy beyond mere bargaining and aggregation of preferences. All suggest that everyone whose vital interests are at stake in a public decision should participate in making it and that true participation requires citizens themselves to engage in discussions with their fellow citizens. Most of these deliberative theorists have examined actual practices of deliberation in an effort to go beyond armchair theorizing. And though they differ in their emphasis, all of these theorists argue that properly conducted deliberation can produce both moral and epistemic advantages. On the one hand, deliberation is supposed to help discern empirical facts and moral truths and (ultimately) lead to the adoption of better public policies. On the other, deliberation is supposed to confer democratic legitimacy. Legitimate, or just, public policies are those that result (or at least could result) from properly conducted deliberation between equals.

This recent scholarship on deliberative democracy is particularly important in the context of the Internet, since scholars have frequently looked at the online public sphere through the lens that deliberative democrats have provided. In Republic.com, for example, Cass Sunstein explicitly evaluates online discourse by the standards of deliberative democracy—particularly through the standards articulated in his own previous work, which reads American constitutionalism as intrinsically deliberative.¹⁹ Sunstein's worry—more prominent in Republic.com than in Infotopia —is that the Internet will fragment the public sphere that deliberative democrats depend on. Deliberation within “echo chambers” will promote polarization rather than respect and democratic legitimacy.

Like Sunstein, Benkler, too, relies on deliberative theorists in his efforts to evaluate the Internet's impact on political freedom. Rather than offer an ideal conception of democracy, Benkler instead asks, “What characteristics of a communications system are sufficiently basic enough to be desired by a wide range of conceptions of democracy?”²⁰ Despite this claim to minimalism, the political philosophers he names directly—Habermas, Ackerman, and Rawls—are nevertheless deliberative democrats of one stripe or another. Benkler singles out Habermas's views several times in order to support his own case. In addition to relying on Habermas for his initial definition of the public sphere, Benkler also echoes Habermas's critique of commercial mass media as reducing public discussion to its lowest common denominator.

Following Benkler and other recent commentators, most of this essay is focused on discussions about blogs and other Web sites that are devoted to political commentary and advocacy. Yet it is worth emphasizing at the outset that a normative defense of blogging is essentially a fallback position. Initially, hopes about the Internet and public discourse centered on new online collaborative forums, where citizens could discuss important public issues, often in real time, in a virtual space. Many such online forums were constructed to reflect the norms and institutional guarantees that deliberative democrats suggested.

Overall, however, reports about these experiments in online deliberation have been dismal. A few scholars reported small-scale successes.²¹ Yet these were the exception. Surveying a large set of diverse experiments in online deliberation, Chadwick harshly concludes, “The road to e-democracy is littered with the burnt-out hulks of failed projects.”²²

If the online forums that have largely failed were designed, from the ground up, to meet both the practical and normative demands of deliberative theorists, the blogosphere was not. The enormous growth of blog readership was a surprise even (as most have acknowledged) to the bloggers who have benefited most. This spontaneous growth and substantial readership is, of course, central to blogging's power. But it also means that the political blogosphere has features that deliberative democrats would never have designed into it.

The Networked Public Sphere: Theory and Practice

Benkler relies on a definition of the public sphere derived from deliberative theory and suggests that these deliberative standards are a good yardstick (though perhaps not the only yardstick) with which to judge the Internet's impact. Yet Benkler is also clear that his vision is not a rehashing of mid-1990s cyberutopianism, explaining that it is silly to dismiss the Internet simply because it failed to make “everyone a pamphleteer.”²³ He argues, quite reasonably, that we should evaluate the Internet against the baseline of commercial mass media. The claim here is thus explicitly comparative, placing online content against the backdrop of traditional media.

Benkler positions his view of the Internet's effects between two distinct criticisms of Internet content. On the one hand, early critics of the Internet worried that chaos and overabundance of content would make it impossible for citizens to gather information effectively. Benkler terms this the “Babel” criticism.²⁴ On the other hand, later scholarship worried about the opposite problem. With winners-take-all patterns in the structure of the Web and in online traffic, the fear was that the Internet would allow a small number of popular outlets to dominate, mirroring patterns found in traditional media.

In contrast to these two prevalent criticisms, Benkler stakes out what he terms the “Goldilocks” position, arguing that the level of concentration seen online is “just right.”²⁵ Benkler's claim is that the networked nature of Web content provides adequate visibility within smaller communities, while also allowing quality content to filter up to a broad audience. At the micro level, Benkler argues, “Clusters of moderately read sites provide platforms for a vastly greater number of speakers than are heard in the mass-media audience.”²⁶ Yet such clusters are egalitarian and not isolated: “As the clusters get small enough, the obscurity of sites participating in the cluster diminishes, while the visibility of superstars remains high, forming a filtering and transmission backbone for universal uptake and local filtering.”²⁷

Benkler's claims here are similar to those made by others. Sunstein's most recent work argues that “the blogosphere might be seen as a kind of gigantic town meeting.”

The presence of many minds is particularly important here. If countless people are maintaining their own blogs, they should be able to act as fact-checkers and as supplemental information sources, not only for one another but for prominent members of the mass media.... The blogosphere enables interested readers to find an astounding range of opinions and facts.²⁸

Drezner and Farrell, in their study of blogs, also emphasize that widely read bloggers promote postings by more obscure bloggers.²⁹ Drezner and Farrell suggest that since journalists themselves focus their readership on top blogging outlets, a story that reaches an A-list site can jump to traditional media. Drezner describes blogger coverage of the CBS-forged document scandal (discussed in more detail later in this essay) as “like firing a flare,” with blogs pointing out the story and traditional journalists investigating and fleshing out the details.³⁰ Journalists and bloggers themselves have also repeated such claims over and over. Bloggers such as Glenn Reynolds³¹ and Hugh Hewitt³² on the right and Jerome Armstrong and Markos Moulitsas Zuniga³³ on the left have all published books arguing that the Internet empowers “an army of Davids” and allows citizens to “crash the gates.”

Yet such egalitarian hopes are problematic, as Benkler's “Goldilocks” account demonstrates. Benkler relies heavily on a piece of scholarship from NEC Research Institute to support his claim that clusters of Web sites are less concentrated—and thus more transparent—at the microlevel. In the article “Winners Don't Take All,” Pennock and others look at clusters of Web sites in a variety of different categories, from online retailers and media Web sites to university Web sites and sites of photographers.³⁴ Benkler cites this research as the most important piece of evidence that we find more egalitarian patterns at the microlevel of the Web. Several other researchers and works cited by Benkler address the same issue but report apparently contradictory findings.³⁵ Benkler's rationale for accepting the ostensible conclusions of Pennock and others over those of other researchers is not made clear.

Given the significance of the research of Pennock and others to Benkler's argument, it is important to note that their claims seem to have been misread. Pennock and others do not contradict the much larger, well-established volume of literature concluding that most communities of Web sites are highly concentrated. Pennock and others model microlevel traffic by assuming that sites get some baseline number of hyper-links just for being part of a community, while another—highly skewed—portion of their hyperlinks depend on their rank within the community.

Pennock and others do find that in some communities of content, sites seem to get a greater portion of links just for belonging. Still, the majority of categories they examine distribute more than 90 percent of their links according to winners-take-all patterns. Moreover, the groups of Web sites that Pennock and others find to be less concentrated are exceptional. The Web sites of professional photographers face natural geographical limits: you cannot hire a wedding photographer from Maine if you are having your ceremony in Florida. Links from one university Web site to another, another apparent exception, are parasitic upon real-world social networks that provide both horizontal and vertical visibility. For example, professors are likely to be aware of scholars at Harvard and Yale and Princeton, as well as scholars at nearby institutions.

Benkler also points to Drezner and Farrel,³⁶ arguing that their research confirms that political content is less concentrated than other online content, making it possible for political Web sites to provide adequate levels of visibility to numerous participants. Yet here again, “less concentrated” is a relative term. Drezner and Farrel are actually making a more technical argument, about whether the distribution of links and traffic among blogs better fits a power law or an extreme lognormal distribution. Even the smaller claim that Drezner and Farrel are making is contradicted by other research that Benkler cites. Adamic and Glance's finding that links among political Weblogs are distributed according to a power law distribution with an exponential cutoff seems difficult to refute.³⁷ Still, whichever claim is right is largely irrelevant to the questions that Benkler raises. Neither set of scholars disputes that political content online is overwhelmingly concentrated.

Yet even if we were inclined to accept the evidence of Pennock and others wholeheartedly, we cannot conclude that all online content automatically provides for “vast” numbers of “moderately read” Web sites. If there is variation in the structure of online content niches, the question is whether the content that Benkler cares about—political content and public discourse generally—looks like the distribution of either retail or photography Web sites. In partnership with researchers from the same NEC Research Institute unit that Benkler relied on, I have conducted research that provides one answer to this question.³⁸ Looking at categories of political Web sites focused on such issues as gun control, abortion, and the U.S. Congress, it does show that some categories of sites are more concentrated than others. Yet even though the study finds hundreds of Web sites in each topical community, the majority of links in each case are divided up between less than a dozen sites. According to the model of Pennock and others, all of these areas of political content look like the winners-take-all patterns found in online retailing and not at all like the more diffuse patterns found among photographers or university home pages.

Traffic Patterns and “Discourse Elitism”

Dividing up Web content into subcategories and sub-subcategories does not demonstrate the conditions Benkler lays out for “universal uptake.” This same conclusion is powerfully reinforced by broad patterns of online traffic. Traffic analysis cannot give the level of detail provided by analyzing link structure. The reason is simple: even with a sample of millions of people, so few people visit the most obscure Web sites that traffic is difficult to measure reliably. This difficulty alone suggests a problem with Benkler's arguments. Still, traffic data gets more directly at concerns about visibility, and it does a better job of placing Benkler's claims about the public sphere in context.

One of the best sources of Web traffic data is Hitwise, an online competitive intelligence service. Hitwise partners with Internet service providers to track and analyze the behavior of their subscribers, allowing for highly detailed, clickstream data. Hitwise's sample includes data from ten million American households, and Hitwise tracks visits to more than eight hundred thousand of the most popular Web sites. Hitwise data thus allows us to look at traffic patterns even within content areas that are a tiny fraction of the Web's total traffic.

Markus Prior asserts earlier in this volume that the Internet and increasing media choice in general has amplified the importance of citizen motivation. Prior argues that citizens with little interest in politics and without firm ideological or partisan commitments have shifted away from political news and toward “soft,” entertainment-driven content. The broad outlines of the Hitwise data are strongly consistent with Prior's argument. News and media Web sites are only a small portion of what citizens see online, accounting for only 3 percent of total Web visits. Online, public oriented content has to compete for attention with countless other materials. If most of what Americans are doing online has nothing to do with issues of public concern, this surely recasts Benkler's argument.

Looking more closely at these numbers raises even more questions. According to Hitwise data, Benkler's claim that Internet audiences are more dispersed than those for traditional media seems to be wrong. Internet audiences may be less concentrated than those for television, particularly with regard to news content. Yet despite growing use of online video, the Web remains overwhelmingly a text-based medium. A majority of the top news and media sites are online outposts of print publications, and even the sites of television news outlets (e.g., CNN or NBC News) distribute far more stories in print form than in video. In measuring audience concentration, print media therefore seems to be a more apt yardstick for comparison. Circulation data is readily available for all major U.S. newspapers and national magazines, courtesy of the Audit Bureau of Circulations, a private oversight organization.

In fact, audiences for online news are both more and less concentrated than audiences for newspapers and magazines.³⁹ Comparing circulation figures to site traffic shows that the most popular news outlets are even more important online than offlline. The top ten media sites receive 29 percent of total site visits, whereas the top ten newspapers and the top ten national magazines receive only 19 percent and 27 percent of total circulation, respectively. More holistic metrics of concentration emphasize the same conclusion. Both the Gini coefficient and Herfindahl-Hirschman Index (HHI), the most commonly used metrics of concentration in the social sciences, suggest that online content is more, not less, concentrated than print media.

Concentration at the top is nevertheless only part of the story. Hitwise tracks more than thirty thousand media sites, and those ranked below 500 account for more than a quarter of the total traffic. Collectively, lower-ranked outlets thus receive far more attention online than they do in any traditional media. The Internet, then, is hollowing out the audience for online news, shifting eyeballs to the most and least popular outlets at once.

Partly, this counterintuitive result seems to come from exposing previously protected local monopolies to nationwide and even worldwide competition. The fate of online newspaper sites illustrates this problem starkly. Not only do the top ten newspapers get dramatically more of the total audience share online than in print, but it is overwhelmingly nationally prominent newspapers (e.g., the New York Times, the Washington Post, and the San Francisco Chronicle) that have a greater online market share; smaller newspapers, by contrast, have lost ground relative to their larger rivals. Controlling for potential confounding factors further strengthens this conclusion. Much content in local papers, for example, is provided by national and international wire services, such as the Associated Press or Reuters. Yet research by Paterson shows that here, too, online content provides less diversity than print media.⁴⁰

Even if online news outlets were a larger portion of the total content and even if top online news outlets were less influential, Benkler's account would still face problems. Most important, the political content that citizens are exposed to on the Internet is still provided overwhelmingly by the commercial mass media. Benkler is kinder to the New York Times, the Wall Street Journal, and the Washington Post— organizations that “may credibly claim to embody highly professional journalism”⁴¹ —than he is to local and regional newspapers. In this respect, shifts from local to national news outlets may be partly positive. Yet given how damning Benkler's critique of commercial media is, sending a proportionally larger number of eyeballs to the most popular news outlets does not seem like a step forward.

The content that Benkler is most interested in, then, is not news and media sites but the sites that Hitwise places in its “politics” category. Most of the specific examples that Benkler uses in his chapters on political freedom belong in this grouping. According to Hitwise, the politics category gets only 0.1 percent of total site traffic, less than one-thirtieth the traffic received by traditional news outlets.

We can argue all we want about what is going on with this one-tenth of 1 percent of all Internet traffic. But no matter how visibility within this niche is distributed, it is hard to see how such a small portion of the public's media diet can account for the positive effects Benkler hopes for. By way of comparison, Hitwise reports (as of February 2006) that 13 percent of all Web traffic goes to sites featuring adult content—outpacing visits to political sites by two orders of magnitude.

Figure 1 graphs the traffic among the top fifty political sites as of February 2006, according to Hitwise data. For that month, Hitwise tracked visits to 970 of the most popular political Web sites. Here, too, we see winners-take-all patterns that contradict Benkler's claim that visibility emerges within small enough categories of content. These top fifty sites account for 62 percent of political traffic. The top ten sites together receive as much traffic as the next forty sites combined. Yet as the figure shows, even the disproportionate amount of traffic received by these top sites understates their importance. Top sites in the community do seem to be functioning as filters, much the way that Benkler and others have suggested. In practice, this magnifies the Influence of these top sites even beyond the disproportionate number of page views they receive. As with news and media sites, it is the “middle class” of outlets that seem to be missing.

Both of these features of online political audiences—concentration at the top and diffusion at the bottom—work against Benkler's claims. Even within the tiny politics niche, top sites have replicated a broadcast-style model of public attention. Moreover, those who get heard in the online public sphere are in many ways less representative and more elite than those whose voices were carried by traditional media. It is common to hear talk of an A-list of bloggers whose voices are disproportionately influential. Yet the top ten or top twenty bloggers who constitute the so-called A-list are only the tip of the iceberg.

This can be seen clearly in my forthcoming research that includes a census of top political bloggers.⁴² Following the widely held belief that the 2004 general election was the moment that bloggers arrived as a political force, the census looked at all political blogs who had more than two thousand daily visitors as of December 2004, according to N. Z. Bear's Blogging Ecosystem project.⁴³ Through public sources and e-mail surveys, data was gathered on seventy-five of the eighty-six blogs that met this level of traffic.

Given that the United States is a nation of three hundred million people, it is striking that so few blogs attracted as much attention as a college newspaper. Nearly all of the bloggers in the sample had careers as lawyers, professors, journalists, senior managers, or technology professionals—hardly a representative cross section of American society. Bloggers were also overwhelmingly white and male and extraordinarily well educated: roughly two-thirds had attended an elite college or university, and 60 percent had graduate education. Perhaps most striking, more than two-thirds of traffic within the sample went to sites of bloggers with a JD, MD, or PhD.

If we compare the thirty op-ed columnists published by the New York Times, the Washington Post, the Wall Street Journal, and the Los Angeles Times over the same period to the top thirty bloggers, the results look similarly dramatic. The bloggers are more likely to have attended a top university; while 20 percent of the op-ed columnists have a doctorate, 75 percent of the bloggers do. Moreover, the columnists provide greater substantive representation of women and ethnic minorities.

In part, the elite profile of top bloggers is good news. Nearly all of the most widely read blogs are published by individuals with professional socialization in journalism, law, or academic research. These are all areas with strict professional ethics against factual inaccuracy. Despite persistent claims that bloggers are a bunch of pajama-clad, rumor-mongering amateurs, the bloggers that actually get read are better educated and better credentialed than traditional opinion journalists.

At the same time, these findings certainly contradict the central thrust of Benkler's account. Benkler declares in reference to the mass media, “a society that depends for its public sphere on a relatively small number of actors... to provide most of the platform of its public sphere, is setting itself up for, at least, a form of discourse elitism.” If the actors are prominent independent bloggers rather than firms, the biases may be different, and commercial pressures may play less of a role. In mainstream media, Benkler asserts, those who are “inside” the media are “able to exert substantially greater Influence over the agenda, the shape of the conversation, and through these the outcomes of public discourse, than other individuals or groups in society.”⁴⁴ Yet as his own examples show, elite bloggers are often able to set the agenda for the blogosphere, and occasionally they are able to set the agenda for mainstream media itself. Benkler's defense of political blogging is a call for egalitarianism. In practice, bloggers are a powerful embodiment of the “discourse elitism” he denounces.

All deliberative theorists emphasize, in different ways, the importance of inclusivity in the public sphere. Yet some theorists have long dissented from the deliberative consensus, arguing that deliberative democracy is really a rebranding of political elitism. As Lynn Sanders puts it, “some citizens are better than others at articulating their views in rational, reasonable terms,” and those whose voices are left out are likely to be disproportionately female, ethnic minorities, and poor.⁴⁵ Peter Berkowitz states the case even more strongly, arguing that the inclusive rhetoric surrounding deliberative democracy is nonsense.

Since it shifts power from the people to the best deliberators among them, deliberative democracy... appears to be in effect an aristocracy of intellectuals. In practice, power is likely to flow to the deans and directors, the professors and pundits, and all those who, by virtue of advanced education, quickness of thought, and fluency of speech can persuade others of their prowess in the high deliberative arts.⁴⁶

Something very much like Berkowitz's vision has already taken hold online. The online public sphere is already a de facto aristocracy dominated by those skilled in the “high deliberative arts.”

The Tail of the Blogosphere: Long, but Not Public

If the political blogosphere has promoted a form of discourse elitism, those not-so-elite sites at the bottom of the pyramid also present problems for Benkler's account. Collectively, these sites may get a substantial fraction of traffic, but any single one of them is insignificant individually. The problem is that the least trafficked sites do not seem to meet Benkler's own definition of what counts as public. And they certainly do not meet the standards of inclusivity or “publicity” endorsed by deliberative theorists.

Conversations about matters of public concern, Benkler emphasizes, are not automatically part of the public sphere. As he puts it, “dinner table conversations, grumblings at a bridge club, or private letters” do not count as part of the public sphere if “they are not later transmitted across the associational boundaries to others who are not part of the family or the bridge club.”⁴⁷ This standard reminds one of the famous Zen koan about a tree falling in the forest. If someone writes about politics on his blog but nobody reads it, does it count as part of the public sphere? Benkler's answer is an unequivocal no.

The problem with this is that 99 percent of Web content about public issues doesn't qualify as part of the public sphere by this measure. Technorati is one of the most widely used indexes of blog content; as of November 2006, it claimed to track more than thirty million “active” blogs (i.e., blogs that were updated within the past three months).⁴⁸ Still, as C. Edwin Baker puts it, “on a typical day... over 99 percent [of these blogs] will be lucky to receive a single visit.”⁴⁹ Similarly, in each of the topical political communities mapped earlier, the large majority of sites only got links in the single digits.

Online discussion is fundamentally different in one respect from conversations over a dinner table or a card table. Most online content is potentially public. As the following examples illustrate, a single link from a widely read outlet can transform an obscure blog posting into front-page news. Yet Benkler's standard suggests that content becomes part of the public sphere only when it is transmitted across associational boundaries—not a moment before. Therefore, those who control the act of transmission have the power in the online public sphere. In other words, audience matters. Not only are most bloggers not public, but they cannot become public without help from their more established colleagues.

Scandal and the Online Public Sphere

Thus far, I have presented evidence that traffic patterns on the Web do not allow for the large corps of moderately read Web sites that Benkler's account requires. I have suggested, too, that the patterns of concentration that do exist have made online discourse less accessible in some ways than traditional media outlets. Online gatekeepers are substantively less representative of the public—in terms of education, gender, and race and ethnicity—than the old media commentators that offer the most direct comparison.

Yet what of instances where the Internet has mattered? Thus far, the Internet's clearest impact on public discourse has been made through the scandals it has either discovered or allowed to unfold more rapidly. These are the sorts of incidents that have been repeated, over and over, in media accounts of why blogging matters. The resignations of Senate majority leader Trent Lott, New York Times editor Howell Raines, and Rep. Mark Foley have all been attributed to blog-generated pressure. The specific examples that Benkler cites, too, follow this pattern. Problems with Diebold voting machines and Sinclair Broadcasting's attempts to air an anti-Kerry documentary on the eve of the 2004 election both generated spontaneous Internet-organized campaigns. (The role of prominent liberal blogs in fueling and coordinating these efforts seems to have been crucial.)

Scandals thus seem to be something of an exception to the claims previously made. Scandals may be one area in which ordinary citizens can be heard online. Yet I argue that scandals nonetheless reveal much about the limits of the online public sphere.

Reporting scandals is an important part of the public sphere's function. Some scholars have suggested that democratic citizenship can function effectively even if citizens mostly just respond to “fire alarms” or “burglar alarms”—in other words, if they pay attention to politics only in the event of a scandal. Schudson calls this “monitorial citizenship” and argues that it is an acceptable—and far more realistic—alternative to informed citizenship.⁵⁰ Other prominent scholars have offered generally approving accounts of this notion as well.⁵¹

Still, it is impossible to ignore the fact that monitorial citizenship is a retreat from deliberative principles. The appeal of monitorial citizenship derives from the fact that it doesn't require the time, energy, attentiveness, or thoughtfulness that traditional republican citizenship demands. To paraphrase Oscar Wilde, monitorial citizenship doesn't take up many evenings. If we're going to evaluate the online public sphere by the standards of deliberative democracy, celebrating monitorial citizenship is simply conceding defeat.

Scandals do not constitute the sort of moral discussion that some theorists take to be a central justification for deliberation.⁵² For one thing, they do not typically involve areas of tough moral disagreement. There was little debate about whether Rep. Mark Foley should have been sexually propositioning underage House pages. No one thought that CBS should base its reporting on forged documents. Partisans and pundits may have disagreed about the precise meaning of Trent Lott's remarks at Senator Strom Thurmond's birthday party, yet both sides loudly repudiated the segregationist ideals that underpinned Thurmond's 1948 presidential campaign. In short, scandals are powerful political moments because they accuse public figures of doing things that the public already agrees are unacceptable.

Scandals are thus unusual. They represent political information of extremely high value, appeal to widely shared political values, and are usually easy to understand. Most often, scandals also involve information that serves the interests of one set of partisans or another. All of these characteristics make scandals exceptionally transmissible within networks.

The fear is that those at the bottom will only get noticed when what they have to say is congruent with the views and interests of the gate-keepers. In part, it is this fear that pushes Benkler to reject commercial media—which he suggests do a fine job of covering issues that will bring them commercial success. It is only when the interests of the media do not align with the interests of the public that problems ensue. Scandals do not test this proposition.

Moreover, the details of the most prominent blog-driven scandals don't seem to include much of a role for ordinary citizens. Consider the incident that right-leaning bloggers branded as “Rathergate.” On September 8, 2004, CBS claimed to have unearthed documents from the Vietnam War era proving that George W. Bush, then a lieutenant in the National Guard, had not completed his service. Late that same night, a posting on the conservative forum FreeRepublic.com claimed that the report was based on forged documents. The apparent forgery was publicized by conservative bloggers and (subsequently) by traditional media. NBC ultimately said that it could not substantiate the documents, firing the report's producer. When Rather retired in March 2005, many suggested that the controversy served to shorten his tenure.⁵³

Initial media coverage suggested that this was a clear instance of the Internet empowering the voiceless such that an online “nobody” could take down the biggest of “big media” Goliaths. Yet when the Los Angeles Times tracked down the pseudonymous poster who had made the initial claim of forgery, they found that “Buckhead” was the alter ego of Harry MacDougald, a prominent Atlanta lawyer and well-known GOP activist.⁵⁴ In the aftermath of the Lewinsky affair, MacDougald had led the national fight to disbar President Clinton over perjury charges.

A remarkably similar story has emerged concerning the Mark Foley scandal. On September 24, 2006, copies of Foley's suggestive e-mails to a male House page of high school age were posted on an obscure blog. These e-mails were linked to by Wonkette (a prominent blog focused on political gossip) on September 27 and broadcast by ABC News's Brian Ross on September 28. Ross had received copies of the e-mails more than a month earlier but did not report on them until after the Wonkette blog highlighted the story. (Subsequent investigations found both more explicitly sexual content and evidence that House Republican leadership had long known about Foley's behavior.) Here, too, the initial narrative suggested that the Internet had allowed an ordinary citizen to take down a congressman. In fact, subsequent coverage revealed that the anonymous blogger who published the e-mails was Lane Hudson, a professional staffer at the Human Rights Campaign, the nation's largest gay and lesbian advocacy group. According to published reports, Hudson initially received copies of the e-mails through his work with the organization but was prevented by his superiors from going to the media with them.⁵⁵

In two of the clearest incidents of blog Influence, then, the story was not about empowering ordinary citizens. On the contrary, the moral seems to be that the Internet gives prominent activists, disgruntled professional staffers, and other existing political elites the means to circumvent long-standing institutional constraints. This may be a good thing for democratic practice, at least in some circumstances. But to conclude from such experiences that the Internet is “democratizing” politics is simply to misunderstand the phenomenon.

Conclusion

It is perhaps unfair of me to criticize a few chapters of Benkler's much larger book and ignore the rest—particularly when the project as a whole is so valuable. Collective action is at the heart of politics, and few accounts of collective action in the information age have the depth and richness of Benkler's. If we want to understand what the Internet means for political life—indeed, why the blogosphere is possible in the first place—Benkler provides myriad insights.

It is therefore important not to overstate my disagreements with Benkler. These center on my claim that collective political debate is different from the many other areas of online collective action that Benkler elucidates so well. For example, in discussing the successes of open-source software, Benkler describes the governance structure of these projects as “meritocracy”—hierarchy, but soft hierarchy. Hierarchy works in software programming in part because code can be judged by essentially objective standards: how fast the code runs, how straightforward it is to understand, how resistant is it to those who would try to break it. In this realm, Benkler's arguments about community production are exactly right, because what we care about is the final software product. In open-source software, grossly unequal contributions are perfectly acceptable; it is fine if most development work is done by a small core group while everyone else just finds and fixes bugs.

Benkler's account works well for other forms of political-minded, community-based production, too. If what we want is accurate, well-written entries in Wikipedia, then these hierarchies of online life are a good thing. The top political bloggers are undoubtedly good at what they do. The content on the top political blogs is consistently smart and factually accurate. Few scholars explain why this is so as well as Benkler does.

Yet well-written, credible blog entries are not all that we care about. Deliberative democrats propose that the public sphere is not a product but a process, and this process requires a level of conversational equality that is missing online. From Habermas onward, the goal has been to provide a public sphere where all citizens have a more equal say. Simply put, that has not happened. Talk about “collaborative filtering” or “meritocracy” cannot paper over the enormous online divide whereby a few dozen educational and professional elites get more attention than the rest of the citizenry combined. In a world of thirty million active blogs, most citizens are more likely to win the lottery than to receive a front-page link on DailyKos or Instapundit.

Deliberative theorists argue that the advantages of deliberation are both moral and epistemic—that the conclusions that deliberation comes to are both reliable and fair. Yet thus far, online political discussion seems to provide more of the epistemic advantages than the moral ones. Deliberative democracy asks a lot of those who participate in the public sphere. Those who run top political outlets are far better trained and equipped to meet deliberative demands than most citizens are. In this sense, the “discourse elitism” that Benkler decries is a partial blessing.

Nonetheless, the central problem with the online public sphere is that it excludes so many citizens. It is bewildering—and darkly humorous—to see white, male bloggers with Ivy League degrees writing about how the Internet is empowering “ordinary citizens.” What they really mean by this is that the Internet is empowering people like themselves.

---

NOTES

1. L. Rainie and J. Horrigan, “Election 2006 Online,” January 17, 2007, http://www.pewinternet.org/pdfs/PIP_Politics_2006.pdf.

2. A. Lenhart and S. Fox, “Bloggers: A Portrait of the Internet's New Storytellers” (Washington, DC: Pew Internet and American Life Project, 2006), http://www.pewinternet.org/pdfs/PIP%20Bloggers%20Report%20July%2019%202006.pdf.

3. C. R. Sunstein, Infotopia (Oxford: Oxford University Press, 2006).

4. R. Rogers, Information Politics on the Web (Cambridge, MA: MIT Press, 2004), 3.

5. A. Chadwick, Internet Politics (Oxford: Oxford University Press, 2006).

6. D. Drezner and H. Farrell, “Web of Influence,” Foreign Policy 145 (2004): 32–40.

7. Y. Benkler, The Wealth of Networks (New Haven, CT: Yale University Press, 2006).

8. See also Prior's essay in this volume.

9. G. R. Chaddock, “Their Clout Rising, Blogs Are Courted by Washington's Elite,” Christian Science Monitor, October 27, 2005; J. Shafer, “Judith Miller's New Excuse,” Slate, March 16, 2006, http://www.slate.com/id/2138161/.

10. N. M. Levey, “Anti-Foley Blogger Speaks Out,” Los Angeles Times, November 10, 2006, A13.

11. H. Yeager, “Blogs, Bias, and 24-Hour News,” Financial Times, September 24, 2006.

12. J. Habermas, The Structural Transformation of the Public Sphere, trans. T. Burger (Cambridge, MA: MIT Press, 1989; originally published in French in 1962); J. Habermas, The Theory of Communicative Action, Vol. 1 (Boston: Beacon, 1984); J. Habermas, Between Facts and Norms, trans. W. Rehg (Cambridge, MA: MIT Press, 1996).

13. J. Rawls, Political Liberalism (New York: Columbia University Press, 2005).

14. J. Cohen, “Deliberation and Democratic Legitimacy,” in Deliberative Democracy: Essays on Reason and Politics, ed. J. Bohman and W. Rehg (Cambridge, MA: MIT Press, 1997), 67–91.

15. C. S. Nino, The Constitution of Deliberative Democracy (New Haven, CT: Yale University Press, 1998).

16. A. Gutmann and D. Thompson, Democracy and Disagreement (Cambridge, MA: Harvard University Press, 1996).

17. J. S. Dryzek, Deliberative Democracy and Beyond (Oxford: Oxford University Press, 2002).

18. B. Ackerman and J. S. Fishkin, Deliberation Day (New Haven, CT: Yale University Press, 2004).

19. C. R. Sunstein, Republic.com (Princeton, NJ: Princeton University Press, 2001).

20. Benkler, The Wealth of Networks, 182.

21. E.g., L. Dahlberg, “The Internet and Democratic Discourse: Exploring the Prospects of Online Deliberative Forums Extending the Public Sphere,” Information, Communication, and Society 4, no. 4 (2001): 615–33.

22. Chadwick, Internet Politics, 102.

23. Benkler, The Wealth of Networks, 177.

24. Ibid., 10.

25. Ibid., 239.

26. Ibid., 242.

27. Ibid., 248.

28. Sunstein, Infotopia, 185–86.

29. D. Drezner and H. Farrell, “The Power and Politics of Blogs” (paper presented at the annual convention of the American Political Science Association, Chicago, September 2–5, 2004).

30. B. von Sternberg, “From Geek to Chic: Blogs Gain Influence,” Star Tribune (Minneapolis), September 22, 2004.

31. G. Reynolds, An Army of Davids: How Markets and Technology Empower Ordinary People to Beat Big Media, Big Government, and Other Goliaths (Nashville, TN: Nelson Current, 2006).

32. H. Hewitt, Blog: Understanding the Information Reformation That's Changing Your World (Nashville, TN: Nelson Business, 2005).

33. J. Armstrong and M. M. Zuniga, Crashing the Gate. Netroots, Grassroots, and the Rise of People-Powered Politics (White River Junction, VT: Chelsea Green, 2006).

34. D. M. Pennock and others, “Winners Don't Take All: Characterizing the Competition for Links on the Web,” Proceedings of the National Academy of Sciences 99, no. 8 (2002): 5207–11.

35. E.g., A.-L. Barabási, Linked: The New Science of Networks (Cambridge, MA: Perseus, 2002); B. A. Huberman, The Laws of the Web: Patterns in the Ecology of Information (Cambridge, MA: MIT Press, 2001).

36. Drezner and Farrell, “The Power and Politics of Blogs.”

37. L. A. Adamic and N. Glance, “The Political Blogosphere and the 2004 U.S. Election: Divided They Blog,” Proceedings of the 3rd International Workshop on Link Discovery (New York: ACM, 2005), 36–43.

38. M. Hindman, K. Tsioutsiouliklis, and J. A. Johnson, “Googlearchy: How a Few Heavily-Linked Sites Dominate Politics on the Web” (paper presented at the annual meeting of the Midwest Political Science Association, Chicago IL, April 15–18, 2003).

39. M. Hindman, The Myth of Digital Democracy (Princeton, NJ: Princeton University Press, forthcoming).

40. C. Paterson, “News Agency Dominance in International News on the Internet,” in Converging Media, Diverging Politics: A Political Economy of the News Media in the United States and Canada, ed. D. Skinner, J. Compton, and M. Gasher (Lantham, MD: Lexington, 2005), 145–64.

41. Benkler, The Wealth of Networks, 204.

42. Hindman, The Myth of Digital Democracy.

43. N. Z. Bear, “Blogging Ecosystem Project,” data aggregated by Site-meter.com, http://truthlaidbear.com/TrafficRanking.php (accessed December 2004).

44. Benkler, The Wealth of Networks, 204.

45. L. M. Sanders, “Against Deliberation,” Political Theory 25, no. 3 (1997): 348. See also I. M. Young, Inclusion and Democracy (Oxford: Oxford University Press, 2002).

46. P. Berkowitz, “The Debating Society,” New Republic, November 26, 1996, 36.

47. Benkler, The Wealth of Networks, 177.

48. BBC News, “Blogosphere Shows Healthy Growth,” November 8, 2006, http://news.bbc.co.uk/2/hi/technology/6129496.stm.

49. C. E. Baker, Media Concentration and Democracy: Why Ownership Matters (New York: Cambridge University Press, 2007).

50. M. Schudson, The Good Citizen: A History of American Civic Life (New York: Free Press, 1998).

51. J. Zaller, “A New Standard of News Quality: Burglar Alarms for the Monitorial Citizen,” Political Communication 20, no. 2 (2003): 109–30; D. Graber, “Mediated Politics and Citizenship in the 21st Century,” Annual Review of Psychology 55 (2004): 545–71. See also Prior's essay in the present volume. But see W. L. Bennett, “The Burglar Alarm That Just Keeps Ringing: A Response to Zaller,” Political Communication 20, no. 2 (2003): 131–38.

52. E.g., Gutmann and Thompson, Democracy and Disagreement.

53. J. Walsh, “Who Killed Dan Rather?” Slate, March 9, 2005, http://archive.salon.com/opinion/feature/2005/03/09/rather/index.html.

54. P. Wallsten, “‘Buckhead,’ Who Said CBS Memos Were Forged, Is a GOP-Linked Attorney,” Seattle Times, September 18, 2004, http://seattletimes.nwsource.com/html/nationworld/2002039080_buckhead18.html.

55. Levey, “Anti-Foley Blogger Speaks Out.”