Sunstein, C. (2001). Chapter 3 in Republic.com [Princeton, NJ: Princeton University Press], "Fragmentation and Cybercascades."

Discussion Leader: FrankLester

Key points and claims

At the beginning of Republic.com, Sunstein poses the hypothetical of a world in which advances in technology mean that consumers have become able to filter out any product, including informational content, that they do not wish to encounter. In this hypothetical, newspapers, magazines, television "stations" and "channels" as we currently know them, and even "news" in general have all become obsolete. "People can decide, in advance and with perfect accuracy, what they will and will not encounter," Sunstein writes. "They can design something very much like a communications universe of their own choosing."

Sunstein uses this hypothetical as a jumping-off point for a discussion of the distinction he wishes to make throughout the book between consumer sovereignty and political sovereignty, and democracy and republicanism. One of his main points is that a climate (which he thinks is the climate this country seems headed toward) in which consumers of information have something approaching perfect control over the information they digest decreases both "shared communications experiences" and "exposure to materials that would not have been chosen in advance but that nonetheless are beneficial, both to the person who is exposed to them and to society at large."

Sunstein's discussion of fragmentation proceeds directly from this assumption about the effects of information filtering. Fragmentation occurs when consumers self-select — both what they read and what they do not read (or see, or listen to). Self-selection encourages the formation of strongly specific and homogeneous subgroups. The formation of these subgroups encourages polarization. Polarization can have many detrimental effects, among them the fostering of hatred and violence. Polarization increases the likelihood of "cybercascades," or information dynamics in which a given point of view becomes more and more widespread as more and more people come to believe it. Polarization also undervalues and underproduces what Sunstein refers to as "solidarity goods," goods such as beneficial or socially valuable information that spread and increase in value the more people consume them.

Sunstein also discusses what he calls "group polarization," in which people are more likely to gravitate toward an extreme point of view the more they self-select into communities where extreme points of view are accepted or encouraged. [DerekHansen: I'm not sure this last sentence is an accurate description of "group polarization" - see my comments below.] Sunstein believes that the Internet encourages this phenomenon, citing studies in which online interactions enhance both anonymity and self-categorization in terms of group identity. [DerekHansen: I think this past sentence should read something like: "Group polarization is even more likely to occur in situations where group membership is made salient (i.e., emphasized) and people have a high degree of anonymity, which is precisely the situation of many Internet communities."] Group polarization suppresses pluralism and diversity of thought, leading to what Sunstein terms the "enclave deliberation" effect, in which like-minded individuals talk to only each other and shut out differing or disagreeing points of view, leading to an echo chamber in which the social argument pool is enriched for points of view that may have been previously marginalized or devalued, but in which positions which may lack larger social merit and which may endanger social stability are valorized.

[DerekHansen: It is important to separate out two different effects discussed by Sunstein related to polarization. The first (a selection issue) is that new technologies allow people to more easily find others that share similar views. In other words, because people can find other like-minded people they will be more likely to interact with those people than with others who have differing views. The second (an in-group effect called "group polarization" by social psychologists) is that within a specific group of somewhat likeminded people there tends to be movement toward a more extreme position in the direction that the group's members were originally inclined. This second effect occurs after groups have already formed. Although separate effects, these two effects combine to strengthen the fragmentation of society.]

FrankLester: Thanks, Derek. Looking over the summary, I agree that I could have been more clear in my definition of group polarization and about the distinction between self-selection and group polarization.]

Sunstein uses a discussion about deliberative opinion polls (in which respondents have been exposed to diverse points of view on issues before they are polled) as an example of a way of using online interactions to reduce polarization. Sunstein's ideal polity is one in which differing points of view have an equal chance to be heard and digested and in which citizens have the opportunity to be exposed to those points of view, whether they like it or not, because exposure to differing points of view nourishes an environment of democratic deliberation.

Critique

Sunstein's discussion is 4 years old and was published before the advent of the weblog as a dominant Internet communication technology [LaurieBuis: Can blogs really be considered a dominant Internet communication technology most of the population still doesn't know what they are?]. [FrankLester: You're right. "Dominant" is probably not the right word to use.] If anything, aspects of Sunstein's view seem prescient. Partisan blogs have in general seemed to increase the prevalence of polarization, cybercascades, and fragmentation by encouraging self-selection of audiences and comments that reinforce their slants. In general, American society since the World Trade Center attacks of 2001 has seemed to drift in the direction of polarization and by extension toward the discouragement and suppression of dissent, not just as a result of social and political change but in many cases as a direct consequence of federal policy. (Sunstein has taken up his point that American political discourse has become more of an echo chamber in papers and books subsequent to Republic.com.)

As both Putnam2001 and Galston1999 point out, an increase in the availability of individual choice can often mean a weakening of the bonds that connect individuals to one another. However, Sunstein does not convincingly make the case that polarization and fragmentation are a necessary consequence of online community formation, although that could be a logical extension of his arguments. If anything, he seems to pay insufficient attention to instances that contradict the thesis that fragmentation and polarization are natural consequences of online community.

Connections

Sunstein's discussion of group polarization and fragmentation ties in with Hogg2001's discussion of depersonalization and group identity. Prototypical members of a group are more liked and gain more rewards than marginal members, and prototypicality is based on, among other things, to what extent the member adheres to group norms and assimilates to the demands of the group in a way that enhances the solidarity and cohesion of the group.

WengerChapter8 argues that the idea of "alignment" (coordinating one's self and one's activity to fit within a larger structure) as a mode of belonging to a community of practice is a double-edged sword. Alignment amplifies an individual's sense of power and vision of what is possible, but it can also disempower an individual by demanding allegiance without question, by forcing one to interpret the demands of the community so literally as to close off other perspectives, and by eliminating the community's ability to see its place in the larger scheme of things — to see the forest for the trees. This seems exactly the set of instances that Sunstein is referring to when he talks about the purpose of a democracy being lost in a society in which consumer choice is the sovereign principle and exposure to diverging viewpoints is limited or seen as irrelevant.

[KathyLee] Here's a web project that tries to diversify--Global Voices∞. From their website:

Discussion

LaurieBuis: I agree with FrankLester that in my mind, Sunstein doesn't make a good enough argument for the idea that polarization and fragmentation are a consequence of online communities. While I realize that this is something that is difficult to empirically prove, I don't think that he tried hard enough. I think his logic was good, but I would like to see some more empirics before I make up my mind...

KathyLee: Sunstein invokes "relative anonymity" (p. 71) as a contributor to group polarization without due discussion (he cites one study which he doesn't really explain). What is relative anonymity, and what's the mechanism by which anonymity increases group polarization?

FrankLester: I think a lot of Sunstein has flaws similar to what KathyLee describes. Part of the problem may be that the book is meant to be read by a "wide" audience and thus does not have the rigorous literature review and bibliographic documentation that we associate with scholarly work. But I think the arguments Sunstein's trying to make suffer from minimal support, and the fact that this is a book meant for mass appeal would not constitute a convincing argument in favor of less evidence.

Nonetheless, his general point seems to be intuitively sound. Maybe that is the scary part--we're largely willing to accept it, without the more rigorous evidence that such a broad claim requires.

CaRichardson: How would you go about collecting this evidence? I think it is interesting to think about instruments for measuring the "variance" of opinions represented on a site as well as the "change" in position over time by users exposed to the site content. These are not easy constructs to measure. Could you construct a web site on a controversial topic that actually resulted in a middle ground consensus? Paul recently published a report on the 2004 elections and the effect of using the Internet to learn about the issues in the campaign. Some of these questions were addressed so he must have tried to measure them somehow. I will attempt to add a link to the pdf of this report to the list of recommended readings. Major findings of the report suggest that people who use the Internet are more likely to be aware of alternative viewpoints than those who do not. However, I'm not sure this addresses how information from the Internet influenced their own voting decisions.