Cass R. Sunstein is the most cited scholar working in constitutional law today: Brian Leiter's 2007 citation analysis (available at http://www.leiterrankings.com) finds nearly twice as many citations to Sunstein (6,180) as to the second-most cited constitutional scholar, Laurence Tribe (3,520). Sunstein's influence stems in part from his fecundity—scores of authored, coauthored, and edited books, as well as hundreds of articles and opinion pieces. While his wide-ranging interests include cloning, democratic theory, environmental regulation, feminism, information and communication technologies, and risk assessment, he is best known for his work on the importance of free speech and deliberation for democracy. Sunstein elaborates a republican vision of the U.S. Constitution as committed to and establishing the conditions for deliberative democracy. Returning for inspiration to debates at the founding of the U.S. Constitution and frequently citing the words of Senator Roger Sherman, Sunstein argues that democracy is not simply the registration of opinions and the representation of interests. To function as more than “consumer sovereignty,” it requires that citizens overcome preferences and reason in public-regarding terms. He has thus argued thoroughly and well for those provisions most likely to contribute to citizens' exposure to multiple perspectives and thereby enhance deliberation. In his most recent work, however, he breaks with this position, viewing political and commercial deliberations as encountering similar problems in aggregating information. More bluntly put, the deliberative democrat appears to have morphed into the neoliberal apologist for consumer sovereignty.

The two books by Sunstein reviewed here begin the same way: “It is some time in the future.” Yet the futures differ. In Republic.com 2.0, despite the fact that the republic is a commercial entity (a dot-com rather than a dot-org), unrestricted freedom of choice in a communications market risks undermining the preconditions of democratic self-governance. In Infotopia, the likelihood that deliberation will produce errors, nonsense, polarization, and extremism indicates the need to turn to the market and new software projects and applications as supplements to or substitutes for deliberation. In the former book, dangers to deliberative democracy arise from the Internet and the market. In the latter one, dangers stem from deliberation itself.

Republic.com 2.0 is a revised version of Sunstein's 2002 book, Republic.com. The new version extends the previous discussion of the fundamental dangers posed by the Internet, dangers for which the author uses the metaphor of the “Daily Me,” a personalized news and media aggregator resulting from perfections in filtering technology (the idea of the “Daily Me” comes originally from Nicholas Negroponte of the MIT Media Lab). Sunstein argues that the massive amounts of information made available by the Internet risk the production of fragmentation and polarization. Because people end up relying on filtering systems to reduce information overload, they enclave themselves into like-minded groups and reduce their exposure to the unwanted and unexpected. New technologies, he explains, “make it easier for people to surround themselves … with like-minded but otherwise isolated others, and to insulate themselves from competing views” (p 63). The problem of the Internet, then, is that it is a medium whose attempts to provide access and interconnectivity lead, perhaps paradoxically, to insularity.

Sunstein's point here twists the more conventional claim that the Internet enables people to find those who share their interests into the suspicion that identification and solidarity can go too far. It is an odd claim to make, though, insofar as it is difficult to see how the Internet leads to more insularity than, say, gated communities, country clubs, churches, and private schools, all of which are explicitly structured so as to protect their members from contact with views and experiences to which they object. Furthermore, anyone who has ever opened his or her e-mail to find ads for Canadian pharmacies, group sex, and penis enlargements (which means anyone who has an e-mail account) will have to agree that filtering systems are a long way from perfect and that unexpected and unwanted encounters occur constantly on the Internet.

Sunstein's worry about insulated groups is that they will become more extreme, a point that he supports with evidence from a Colorado experiment on group deliberation and a couple of studies of linking behavior on liberal and conservative blogs (he refers to no content analyses of blogs or surveys of blog readers or writers; additionally, all his examples are from political blogs in the United States, a tiny portion of the blogosphere, which gains and loses thousands of blogs a day). As the author explains, the Colorado experiment demonstrates that when people who share an initial view deliberate, they come to hold their positions with more conviction. Groups with an initial orientation become more extreme; their views become more homogeneous and unified. He applies this insight to blogs, arguing that insofar as they are sites where the like-minded congregate, they similarly risk turning into “echo chambers.” As a blogger, I find this unconvincing; in my experience, commenters can turn the most innocuous statements into opportunities for disagreement, disparagement, and, if we are lucky, wide-ranging flame wars. Put more generally, the appearance of blogs as echo chambers is a matter of perspective: At what level is one looking for and finding similarity and agreement? What looks the same from one angle—the blogger and commenters are all left wing—unravels into a multitude of differences from another—some are socialists, others are anarchists, and still others want to work within the Democratic Party.

Sunstein is concerned that echo chambers lead to extremism and hurt the cause of truth. Although he acknowledges the benefits that groups accrue from deliberation among their members, what he calls “enclave deliberation,” he holds that “whenever group discussion tends to lead people to more strongly held versions of the same view with which they began, there is legitimate reason for concern” (p. 78). But why should we worry—rather than celebrate—that political deliberation leads to politics, that is, to division? Even as he admits that extreme views might sometimes be better than moderate ones, Sunstein remains suspicious of extremism and polarization. This suspicion is misplaced. In a situation of division and disagreement, polarization is the truth of a situation. To bring clarity to division, to make visible the sites, issues, concerns, and values that separate us from them, is politically crucial as it enables differentiation, party formation, choice, and decision. The most significant questions for politics are those of peace and war, life and death, rights and responsibilities, haves and have-nots. To expect moderation on such questions, and to fear the emergence of division in the course of deliberation on them, is to attempt to suppress or displace politics itself. When a deliberative democrat takes such a position “from the standpoint of democracy,” he erases democracy's potentially radical core.

This erasure occurs more fully in Infotopia. Here, deliberation is the problem to be solved, not the supposition of a republic in need of protection. Sunstein's examples slide between corporations, government institutions, and experiments in deliberation. In this future, there is no deliberative democracy. Infotopia emphasizes the problems accompanying deliberation (all of which are mentioned in Republic.com 2.0, sometimes with the same examples and citations, occasionally in the same words, and some of which are discussed in similar terms in Why Societies Need Dissent, 2003): polarization, extremism, confidence, the amplification of errors, and cascade effects.

Cascade effects refer to the way the choices or expressions of some people impact the choices or expressions of others. Sunstein uses speaker order as an example. If the first several speakers agree, more people become likely to “jump on the bandwagon.” Subsequent speakers then take on social risks in disagreeing. The presence of a bandwagon, or the supposition of an emerging consensus, may thus stifle contrary views or conflicting information. Stock-market bubbles, fads, and urban myths all exemplify cascade effects. The author introduces all these points in each book. Yet in Republic.com 2.0 he associates cascade effects with the problems of fragmentation. In Infotopia, he interprets cascades as a problem of deliberation per se (in the earlier Why Societies Need Dissent, he views cascades as a problem of conformity).

As one possible remedy for the problem of cascade effect in particular and deliberation in general, Sunstein suggests prediction markets. A prediction market is a pool of bets on an outcome; because people are more likely to be accurate if money is at stake, the argument goes, they are only likely to risk their money if they are relatively confident in a given outcome. He draws evidence for the accuracy of prediction markets from business settings, where prediction markets have been used to determine product launch dates, prices, and store openings, as well as from politics, where prediction markets on electoral outcomes have been used in forecasting. Although he is careful to acknowledge specific failings of prediction markets, he evinces nonetheless an underlying faith in the power of markets to generate sensible outcomes, on the grounds that people have a monetary stake in them. Perhaps because I am writing this review during a colossal crash in housing prices resulting from the collapse of the subprime mortgage market, a crisis that comes only a few years after a series of corporate accounting scandals, which come on the heels of the bursting of the dot-com bubble, I find Sunstein's confidence in prediction markets difficult to take seriously. Or perhaps I was persuaded by his criticisms of a sovereignty of consumers, criticisms that emphasize the difference between self-interest and public mindedness, and arguments, however limited, on behalf of the sovereignty of politics in Republic.com 2.0.

But there is another reason for my skepticism, namely, Sunstein's association of deliberation with the aggregation of information. The argument of Infotopia is premised on the idea that deliberation is a way to release information from many minds so that such information might be shared. The author acknowledges that deliberation over information is not the same as deliberation over values; yet he sees these as matters of degree—deliberation over values is like deliberation over facts; it's just harder. Indeed, it is only because of this assumption that he can discuss software applications like blogs and wikis and open-source software projects as analogous to deliberation; each is in some way an effort to produce a specific outcome. But this is no more convincing than the idea that answering a question the answer to which is unknown to some but uncontested as an answer (like on a game show) is not fundamentally different from the problems of politics—who are we and what should we do? Open-source software projects are poor models for political deliberation because their orientations and outcomes are clear—bug-free software. Bug-free software is not the same as, say, peace in the Middle East. Blogs and wikis cannot substitute for political deliberation because they are only software applications, tools for carrying out a project the value of which is already accepted by the participants or platforms that might be used to express or debate an idea without having to take responsibility for reaching and executing a decision. They are not themselves exchanges of reasons or opinions (although they are constituted through such exchanges).

Infotopia, like Republic.com 2.0, ends with a few policy suggestions. What is striking, though, is that its suggestions are for business and government alike, as if their interests were the same. Republic.com 2.0 at least acknowledges the conflict between them, the way that the decisions consumers make may well fail to consider larger, collective interests and ideals.

As I read these books, I came away thinking that Infotopia was closer to Sunstein's actual position, that his neoliberal, Hayek-loving, Chicago-school self had deliberated with his Madisonian, democratic, political self and carried the day. He writes optimistically, and these days it is hard to be an optimistic deliberative democrat. Yet I cannot help but wonder whether he would come to these views had he asked different questions. For example, he does not attend to recent work in network science, particularly that of Albert-Laszlo Barabasi on scale-free networks.

Barabasi uncovers structural features of complex networks, namely, that they are governed by laws of growth and preferential attachment that account for the presence of hubs and the 80/20 or power-law distribution of links. Together, these laws mean that those who come first have more influence: People are more likely to choose something that another has already chosen because it has been validated. They also mean that the difference between the top two positions in a network (for example, a citation network like the one in constitutional law) is significantly greater than the positions between, say, the eighty-third and eighty-fourth positions, or that roughly 20% of the people in a group do most of the work or have most of the money or are cited most of the time. Barbarasi explains why and how this is the case in his 2002 book, Linked. I introduce it here to say that one can recognize the structural inequalities reproduced on the Internet—and in any citation network—without jumping to fragmentation, polarization, and extremism as specific to either the Internet or groups, or to deliberation. Attention to network structures helps dissipate excessive anxiety or optimism around the Internet—the likelihood of any given blog linking to mine in a universe of more than 70 million blogs is close to zero (this indicates as well a methodological weakness accompanying a focus on linking). The presence of extreme views or groups of like-minded blogs tells us little to nothing about their political impact.

More broadly, Barabasi's work on network structure suggests that the problem Sunstein attributes to deliberation and seeks to avoid by turning to markets (better access to information held in many minds) occurs in any complex system and is thus not a fault of informational cascades or group polarization brought about by the discussion. And maybe it reintroduces the importance of politics: If the underlying structure is fundamentally unequal, then changing it will require, in the words of Roger Sherman that Sunstein so frequently quotes, action “for the benefit of the whole community.”