Republic.com 2.0

• Introduction
• Remarks
• Questions and Answers

JOANNE MYERS: Good morning. I'm Joanne Myers, Director of Public Affairs Programs. I would like to thank you all for joining us today. I hope you all had a wonderful summer and are looking forward to the fall and the many exciting programs we have planned for you.

To launch this new season, I'm extremely pleased to welcome back Cass Sunstein. Professor Sunstein first spoke here three years ago. At that time, he talked about why societies need dissent. For those of you who were not here at the time, you can find the transcript of his remarks here.

Cass, as many of you know, is one of our country's preeminent legal scholars. He is a Karl Llewellyn Professor of Jurisprudence at the University of Chicago Law School and also teaches in the Department of Political Science.

I believe you all should have received a copy of his bio. In reading it, you will note that, in addition to this most recent publication, he is the author of numerous other books and articles, writing them almost in the time it takes many people to read them.

Today he will be discussing Republic.com 2.0, which is clearly the result of the time he spent thinking about how to respond to the many questions raised by others in reference to his earlier book, Republic.com. In this instance, he takes a fresh look at both the impact of Web 2.0 and the consequences of the rise of endless communications options, which have raised new issues.

By the way, I just want to point out that "Web 2.0" is a phrase that was coined a couple of years ago with regard to a perceived second generation of web-based communities and hosted services, such as social networking sites and blogs, which now number more than 70 million globally, and wikis, which facilitate collaboration and allow users to share links, videos, and other information with online friends. Although the term suggests a new version of the World Wide Web, it does not refer to an updating of web technical specifications, but rather to the changes in the way software developers and end users use the web as a platform.

In Professor Sunstein's earlier work, Republic.com, he argued that the Internet might weaken democracy because it allows citizens to isolate themselves within groups that share their own views and experiences, thus cutting themselves off from any information that might challenge their beliefs, a phenomenon which we now know as "cyber balkanization."

On this occasion, he develops his thesis further and persuasively argues about the issue that concerns him most, which is how to avoid information cocoons which shut us off from topics and opinions that we would prefer to avoid. To this end, he suggests what we, as consumers and producers, can and should do.

There is no question that in this millennium traditional media has been replaced by a more personalized one, known as the Internet. Rather than using it to seek news, information, or culture, more often we use it to actually be the news, the information, and the culture. Whether we blog, watch amateur broadcasters on YouTube, or enter virtual universes such as Second Life or calling card emporiums such as MySpace and Facebook, the Internet has become a mirror unto ourselves.

As the Internet revolutionizes society, we can't possibly imagine how far or in what direction it will evolve. Yet one thing is certain, as Professor Sunstein will argue: We must ensure that the vast unrestricted choices made possible by technology do not undermine democracy, for only then can we all realize the promise that the Internet holds.

Please join me in welcoming one of our country's most thoughtful and respected legal scholars, Cass Sunstein.

CASS SUNSTEIN: Thank you for coming, early in the morning.

These remarks will be about a conversation I am imagining between Google and John Stuart Mill.

Google begins the conversation: "No one can read all the news that's published every day, so why not set up your page to show you the news that best represents your interests?"

That's how Google starts its celebration of Google News, to which Mill replies, "It is hardly possible to overstate the value in the present state of human improvement of placing people in contact with other persons dissimilar to themselves and in contact with modes of thought and action unlike those with which they are familiar."

"Such communication has always been, and is peculiarly in the present age," Mill says, "one of the primary sources of progress."

As I hope is clear from this imaginary dialogue, my topic is personalization or customization, which is what Google is excited about. The New York Times now has something called "My Times" that you can personalize. This was prophesied by a technological prophet named Nicholas Negroponte in the early 1990s, who referred to the emergence of the "Daily Me," by which, he said—and it seemed too imaginative at the time—people would be able to construct, and actually would be constructing, an informational universe which fit distinctly with their own interests. Negroponte thought that was just great.

More recently, there is a book by the editor-in-chief of Wired magazine called The Long Tail. Do you know this book? What Chris Anderson says in this very good book is that there is now a "long tail" of goods on Netflix or on Amazon. This allows phenomenal niche marketing, by which things on the long tail—meaning the distribution of goods that hardly anyone but a small group wants—will be nonetheless profitable to sell, because in a massive market, at least some people will get that obscure book I wrote on constitutional law in 1994.

So what Anderson celebrates is the "nichification" of the market, saying this is now profitable, and the future.

The rise of the 70 million blogs—actually, now it's 71 million; in the last ten minutes, a million were added—is part of the general phenomenon.

The long tail, the "Daily Me," and the rise of the blogosphere are meant to signal a utopian vision, which Google clearly is excited about. And it is, in many ways, terrific, probably in more ways terrific than not terrific. But I am going to try to draw attention to an unfortunate side. The inspiration is a series of studies I have been involved in that are not directly about the Internet. I'm just going to tell you about two right now.

The first involves Colorado. We tried to construct a little bit of a "Daily Me" in Colorado, in the pretty recent past, by having the people of Boulder, Colorado, who are liberal, get together and talk to each other about climate change, same-sex civil unions, and affirmative action. We thought the people in Boulder were liberal, because it goes about 70 percent for the Democratic candidate in presidential elections, but we wanted to make sure. So we asked them a couple of screening questions, such as, "What do you think of Vice President Cheney?" If they said, "He's okay," we excused them from the experiment.

So we got liberals in Boulder to talk, and on the same day we had people in Colorado Springs, which is Bush country, talking about exactly the same issues. We excused them if they were skeptical about Vice President Cheney. We wanted them to like Vice President Cheney—and sure enough, they did—if they were to be included in the experiment.

Here's what happened. We had them deliberate together on these three issues.

The median view before deliberation, in Boulder, was pretty positive about an international treaty to control greenhouse gases. After a period of discussion with one another, they were really, really enthusiastic about an international treaty to control greenhouse gases.

The people in Boulder were somewhat diverse on affirmative action, though basically the median person liked it, before they talked. After they talked, they were very enthusiastic about affirmative action.

They got more extreme than they were before on all three issues. At the same time, internal diversity was dramatically reduced in their anonymous post-deliberation statements of view.

So in Boulder, the citizens were liberal, but there was diversity on same-sex civil unions before they talked. After they talked, in their anonymous post-deliberation statements of view, they were more liberal and they were much less diverse. So they got more extreme and they got much more uniform.

The exact same thing happened, moving to the right, in Colorado Springs. Diversity was squelched; extremism was promoted.

As a result of the Boulder people going to the left, the Colorado Springs people going to the right, where they were separated by this amount in terms of their median views (anonymous) before deliberation, afterwards their median views (anonymous) were this far apart. They got dramatically separated.

The only other study I am going to tell you about involves federal judges. In a way, the federal judges have been conducting their own version of the Colorado experiment, with some overlap with the blogosphere. Here is the way in which they have been doing that—not through their own intentional actions, but just through the construction of federal panels.

Court of appeals panels in the United States are three judges. They are randomly assigned. They consist—and these are the only possibilities—of three Republican appointees, three Democratic appointees, two Republicans and one Democrat, two Democrats and one Republican. That is the only possible composition in federal courts of appeals.

A few years ago, I asked a research assistant of mine who had nothing to do that week to look at some environmental cases to see how Republican appointees vote, depending on whether they are sitting with two Republican appointees or at least one Democratic appointee. If we construct something like Colorado Springs, Bush country, on the federal judiciary, just by looking at RRR panels, how do RRR panels look in environmental cases compared to how they look when it's mixed?

She collected about 40 or 50 votes. We didn't have enough to do statistical tests, but we did have enough to be startled, to find that Rs, Republican appointees, show very conservative voting patterns on the federal courts when they are sitting with two other Rs. In a case in which the Sierra Club or the Natural Resources Defense Council is suing the EPA [Environmental Protection Agency] to get it to do more, Rs on our panels vote for them about 20 percent of the time. Rs are much more likely to vote for them—environmental groups—when there is at least one D present. The divergence is very dramatic.

Then I asked her to check it out with Democratic appointees to see if she could find the same thing. Sure enough, she did. Ds in gay-rights cases vote for the plaintiff, the last time we checked, on DDD panels 100 percent of the time. Rs on RRR panels vote for gay-rights plaintiffs 14 percent of the time. That's a massive difference. That is in line with what we generally observe.

What the Colorado experiment and what the judicial data show is that likeminded people who are talking to one another end up in a more extreme position, in line with their pre-deliberation tendency. There is a shift toward extremes and toward uniformity as a result of deliberation among the likeminded.

So what I am going to be emphasizing, in light of that risk, is the importance for democratic self-government of unanticipated, unchosen encounters with diverse ideas and subjects. In other words, I'm going to be taking Mill's side against Google.

The way I am going to try to get at this is through obscure and somewhat exotic constitutional doctrine, called the public forum doctrine, whose impact you will notice in New York City in the next week if you watch for it, notwithstanding the obscurity of the doctrine.

The free-speech principle is mostly thought to be about censorship, and it is. That is its core goal, to stop censorship. But it's also about the construction of spaces in which diverse people can get together and reach a heterogeneous audience. Every tyrant knows, in Cuba and China and Saudi Arabia, that the anti-censorship principle is important to overcome and to censor, but it's also important to close off areas in which people might congregate for expressive activity.

If we look at the public forums that the Constitution protects, we will see that they are the streets and parks, which, under the exotic constitutional doctrine, have to be open for expressive activity. The streets and parks were the big public forums of the 18th and 19th centuries, in which people had a chance to protest specific misconduct, as they saw it, if the street or park was near the alleged wrongdoer. They had a chance also to reach a diverse public, insofar as the streets and parks had a degree of diversity, as they certainly do in Chicago and New York. Also the public forum doctrine, insofar as it was about streets and parks, imposed a kind of duty—not a legally enforced one; kind of a practical one—on each of us to see the diversity of concerns of our fellow citizens and to encounter, once in a while, them and their objections to what was happening. So the public forum doctrine, insofar as it was salient, prevented us from living in gated communities—"us" meaning whatever our demographic or religious or racial characteristics. Gated communities became difficult.

In the 20th century, the general-interest intermediaries—newspapers, magazines, and TV stations—provided a lot of this function, and they continue to do so, to a large degree. The idea is that what I am calling the general-interest intermediaries, in contrast to the blogosphere, provide a kind of social glue, and also ensure that some of the time you will have unanticipated and unchosen—and maybe even somewhat unwanted—encounters with topics and points of view that will affect you. So you might think you have absolutely no interest in what is happening in Nairobi, but a story about Nairobi might catch your eye and alter your thinking, maybe alter your interest, maybe even affect your behavior, in a way that, for most people, doesn't happen often, but in the social aggregate, is important.

Some people may not think they like reading George Will very much, but their eyes might encounter a George Will column, or David Brooks or Paul Krugman. If their eyes hit on the story, it may catch their attention and they might learn something, even if they never would have put Will or Brooks or Krugman in their "Daily Me," and even if the long tail of Amazon or Netflix would never have selected these authors for them.

You can see in this light, I think, a problem with something that is getting a lot of celebration these days, which is the phenomenon of collaborative filtering, which says, if you like X, you are going to like Y, and the reason is, people like you who like X also like Y. So if you like Paul Krugman, chances are you will like Anthony Lewis; if you like George Will, chances are you will like William F. Buckley, Jr. The concern is that this constructs something like Boulder or Colorado Springs in our experiment, by free choice.

If we now investigate just a little more the idea of a "Daily Me" or a long tail or "MyGoogle.com," we can see that there are two things that are lost. First is exposure to topics that are important, maybe, for people to see, even if they wouldn't select them; second is exposure to points of view.

We know enough to know that on the Internet, as in Colorado and on the federal judiciary, likeminded people speaking with one another end up thinking a more extreme version of what they originally thought. I got a vivid example of this that I will tell you about in an early discussion of Republic.com, the original book.

There is a Web site called freerepublic.com, which is a libertarian Web site. It still exists, I'm pretty sure. They started talking about this book of mine. They didn't like it very much. As they talked, they hated it more and more, until, finally, the worst insult they could imagine was delivered, which was, "He's married to Greta Van Susteren." By their lights, that's the worst thing you could possibly say. I'm not, incidentally, married to Greta Van Susteren, but that was really—to them this is a satanic person.

Why is it that in Colorado and on the federal judiciary, we see people ending up in a more extreme position as a result of internal deliberation? To get hold of that question, let's pause a little bit over the generality of the phenomenon.

We know now that when white people who are inclined to show racial prejudice talk to one another, they become more inclined to show racial prejudice. Racism breeds its own exaggeration. We know also that white people inclined not to show racial prejudice, who show a little bit of racial prejudice but not much—after they talk to one another, the racial prejudice diminishes to the point where it nearly vanishes.

We know that people, with respect to risk-taking behavior—and this includes burglars—in ordinary circumstances, where entrepreneurs will take risks—maybe think Enron, possibly think White House and Iraq War—where people are inclined to risk-taking behavior as a result of deliberation, what happens is just like what happened in Colorado. They are more inclined to take risks and they are more uniform in their enthusiasm for risk-taking behavior.

We know, with respect to feminism, that women who are inclined to be pro-feminist, after talking to one another become more clear that feminism is correct and that sex-based injustice is pervasive. Women who are hostile to feminism, as a result of internal discussions, show exactly the pattern we observed in Colorado.

There seem to be three mechanisms that account for group polarization. That's the name for the phenomenon.

One is, if you have a group that has a pre-deliberation tendency—as, for example, the participants in the blog, the Daily Kos, if you know it—if they have a pre-deliberation tendency, the arguments offered that will favor that tendency will be more numerous than the arguments that go the other way. So if you have a group of people who believe that the United States should sign the Kyoto Protocol tomorrow, then the arguments that favor that step will, in the relevant group, be larger than the arguments that go the other way. If people are listening to one another—which, thank goodness, they do—they will end up being more extreme in their commitment to the Kyoto Protocol after speaking.

This suggests that the use of the Internet for political recruiting and political discussion will look a lot like what we have observed in Colorado.

There is another mechanism that is a bit of a cousin of this argument-pool mechanism. It turns out that most people in isolation, on issues that aren't terribly familiar—like climate change to most Americans; they are not sure exactly how to think about it—go to the middle. Because they are not that confident in their view, they avoid extremes. They may think, "Yes, we should do something," "No, we shouldn't," but there tends to be a degree of tentativeness.

It turns out that once people find their initial tendency corroborated, they become more confident, and confidence breeds greater extremism. So with respect to a wide range of issues involving politics, the attractiveness of people in slides, the comfortableness of chairs, the desirability of products, corroboration breeds more intensity of commitment. That seems to be the second mechanism.

When I was first finding this group polarization with respect to deliberating groups, I mentioned it to someone at a university, which shall go nameless—a person who shall go nameless—who is an enthusiast for animal rights. The reason I am not going to quote him you will see in a moment. He—and I share his view—believes that animal suffering is pervasive and unjust, and much should be done to reduce it.

He said, "You know, here's what happens in animal-rights groups. When we meet on Friday of a three-day meeting, we are, by my lights, extremely sensible and right. On Sunday, we've lost our minds." [Laughter]

He said, "What happens is, on Friday, we have a bunch of moral commitments and policy positions which are sensible. On Sunday, we believe that no experiment on animals ever delivered benefits for human beings."

The mechanism that explains this isn't quite captured by what I've discussed so far. It is that human beings tend to want to conceive of themselves in a certain way and also to present themselves in a certain way. Research suggests that it's not right to think that people want to be like other people. They don't. People want to be different from other people, but in the right direction and to the right degree. Animal-rights people want to be more enthusiastic about animal rights than other people, and if they find themselves in a group of people who are as committed to animal rights as they, they shift a bit in order to hold on to their preferred self-conception and self-presentation.

I have seen some of the tapes of our deliberations in Colorado, and that's exactly what happened. There is no question that on the Internet, when people are sorting themselves into a community of, let's say, supporters of the Iraq war or critics of Senator Obama, something like what happens in animal-rights organizations happens, too, where there is a shift to the right or the left, depending on how people understand themselves in relationship to the group.

We can understand, in light of these mechanisms—the corroboration-breeds-extremism, the argument-pool mechanism, and the self-presentation/self-conception—something about how ethnification occurs, both in countries that aren't terribly ethnified at some points, as in the United States, or where there is an ethnic movement among certain racial and religious groups. That happens as a result of mechanisms of this sort.

We might also understand something about the wellsprings of terrorism, which is not a product, it turns out, of poverty or lack of education or mental illness. Terrorists are not poorer or less educated or more anguished, as far as the professionals can tell, than similarly situated people. In fact, they are a little less poor and a little better educated. But what they are is subject to echo chambers or polarization machines, often self-consciously created by what we might call polarization entrepreneurs, of which Osama bin Laden is an example.

The Internet is breeding hate groups. Hate groups are quite aware, intuitively, of the mechanisms I have discussed, and they take advantage of those mechanisms, to the extent that people are sorting themselves into communities of very angry types—sometimes acting out, usually not. The Internet is contributing to that.

Let me say a few words about blogs. We have seen in the last few years a remarkable rise in their number and influence. We have also seen a great deal of celebration of blogs as correcting errors of the mass media, and bringing issues to the attention of the public that might be ignored. These celebrations have reason. That is, blogs do operate often to correct error and to aggregate widely dispersed information, in a way that is good for democracy.

The fear, which should be transparent, is that the blogosphere is operating a bit like a spontaneous example of the Colorado experiment, in which people are congregating around conservative or liberal blogs in a way that aggravates preexisting tendencies.

We are starting to get some data on this. I can tell you a little bit about what the data shows. It shows three things.

First, the blog rolls on the conservative blogs typically refer either exclusively or dominantly to conservative blogs. The blog rolls on the liberal blogs show the same skew.

With respect to linking behavior, there is a little more complexity. That is, with respect to linking behavior, it's not the case that the liberal blogs only link to liberal blogs and conservatives only to conservative blogs. There is dominance of strong majority linking to likeminded types, but hardly 100 percent.

The bad news is that when the liberal blogs link to the conservative blogs, it's typically to show how ridiculous and contemptuous the alternative view is, in the nature of, "To see a truly absurd argument and the depths to which they have sunk, see this," which is not a reflection of respectful exchange.

Reading shows, so far as we know, broadly a similar pattern, where the readers of left-wing blogs tend to concentrate on those and those of right-wing blogs tend to concentrate on those.

What ought to be done about this? It's important to say that some problems lack solutions. It is not only the case that, on balance, the Internet is a friend of democracy, not an enemy, but also to say that risks that come from associational liberty maybe are best unaddressed, that they are a byproduct of a desirable technology. Cell phones and televisions and BlackBerries all have downsides, too, but those probably are not best addressed by government.

On the other hand, there is a great deal that the private sector can do. I have just three little recommendations for what the private sector might do.

The first is to have much more in the way of respectful linking, in a way that reflects a tip of the hat or acknowledgement of the good faith and legitimacy of an opposing position. The linking behavior might be thought to reflect a virtue—let's give it a name—called political charity, which is to recognize that the views of those with whom we have intense disagreements are typically motivated by good faith and often have plausible, and maybe even reasonable, grounds.

So my suggestion is, a norm ought to develop by which the National Review, for example, links to The New Republic or Dissent links to The Weekly Standard, and vice versa. It ought to be done by a kind of mutual agreement. If it isn't, then it would be really great if, for example, The New Republic or Dissent just voluntarily agreed to link to The Weekly Standard or the National Review, and see if the National Review or The Weekly Standard would respond in kind. Fine, either way.

The second idea is that we ought to be seeing on the Internet much more self-conscious efforts to create, let's call them deliberative forums in which opposing points of view are engaging with one another in a way that lots of different people can consult if they want to see that sort of thing. There are relatively few, considering how late we are in the Internet game, experiments in deliberative forums in the United States. If some of you have ideas about how to do that, great. Given the technology, it's really easy to do. There are such experiments arising in other countries, sometimes with modest public subsidies, sometimes purely private, always with private initiatives, so far as I am aware.

The more general idea is to try to replicate in the Internet era something akin to the public forum doctrine of the 18th and 19th centuries, something like the equivalent of the streets and parks, and something like the aspirations, at least, of the general-interest intermediaries: the newspapers and magazines and evening newspapers.

I have one story for you and a quotation from John Dewey, the 20th century's Mill, in many ways. Here is the story.

There is a political scientist at Stanford named Fishkin who has been long interested in the problems with opinion polls. He thinks that if you just cold-call people and ask them, "What do you think of X," and then aggregate their views and think the American people think X, then that's really unhelpful. So what Fishkin has done, instead, is to create what he calls a deliberative opinion poll, which is a nice contrast with the "Daily Me," in the sense that the deliberative opinion poll gets diverse people together in a place, asks them to talk to one another about issues, having been presented with material, and then sees what they think afterwards.

Here is what happened in one of his deliberative opinion polls. There was a small group, about six people, talking about family policy. The group had as two of its members a farmer from Oklahoma and a welfare mother from New York City. The woman, who was a single mother of two kids on welfare, was talking about her family and its needs, and the Oklahoma farmer was getting increasingly agitated. People in the group didn't know what exactly the agitation was about: Did he disagree with her? Was he racist? What was going on?

Finally, he exploded at her and said, "In the United States, a family means a mother, a father, and some kids. There's no father in the picture in your situation. Do not speak in my presence of the word 'family.' You do not have a family."

As you can imagine, the conversation for the next couple of days was difficult. In the little group, they did talk, but the Oklahoma guy and the New York woman never spoke to each other directly. They didn't exchange a single word. But they did talk, both of them, about what they thought, just not to each other.

As they left on Sunday, she felt a tap on her shoulder and turned around, and there was the Oklahoma guy looming—she wasn't big; he was—looming over her.

She said, "Yeah?"

"What are the three most important words in the English language," he said, somewhat sternly.

She said, "I don't know."

He said, "I was wrong."

The quotation is from Dewey: "The belief that thought and its communication are now free simply because legal restrictions have been done away with is absurd. The currency of that thought perpetuates the infantile state of social knowledge, for it blurs recognition of our central need, which is to have conceptions that are used as tools of inquiry and that are tested, corrected, and caused to grow in actual use. No one and no mind was ever freed merely by being left alone."

Thanks.

JOANNE MYERS: Thank you very much, Cass, for that wonderful discussion.

I would like to open the floor to questions.

QUESTION: Thank you for a very clear and substantial analysis of an existing problem that's bound to get worse. I have two questions, one on the positive side.

One would hope that universities would be the best place to encourage diverse exchanges, even though some universities are already identified as red or blue. But this should be the whole idea of the university, to have discussion and read different authors and keep the dialogue going.

The second question is where we might be going. If you identified groups of people who are more and more attracted to terrorism and who reinforce each other and who are willing to take action, does this mean that the government, on behalf of all of us our national security and so forth—is going to intervene more and more in order to control these terrorist groups?

CASS SUNSTEIN: Thank you for the kind words and also the great questions.

There is no question that the university ideal—and Dewey was an educator above all—is understood as a way of correcting this. Political correctness, in all its guises, is an exemplification of group polarization. If you get a bunch of left-wing people together on campus, typically they will end up in a more extreme position, in line with their pre-deliberation tendency. I will just say, conservatives have been quite right to object to that as a very unfortunate thing for education and, in the end, for democracy.

It doesn't happen only on the left. I am at the University of Chicago, and our economics department has long been both exceptionally good and a polarization machine. So it's a problem on both sides.

Brigham Young is one example. Berkeley at least used to be, on the other side, another example.

There is an implication for terrorism policy, isn't there? Both President Clinton and President Bush have occasionally said that deprivation is a source of terrorism; poverty breeds terrorism and the lack of education breeds terrorism.

The first is clearly wrong. Poverty is not a source of terrorism, so far as anyone can show. There is no showing that terrorists themselves are poor nor that terrorists come from nations that are poor. The data is very weak on that.

With education, it's mostly wrong, but there is a little grain of truth to it in a way that connects with what you said. The wrong kind of education does breed terrorism, and the right kind of education does correct against terrorism. So if you have the kind of education which I gather is occurring under the Palestinian Authority, which is educating kids, even at a very young age, to the value of suicide bombing and such, that has a large impact.

So the antiterrorism policy ought not to be focused on relief of deprivation or on getting schooling. It ought to be concerned about disrupting networks by which terrorism is bred. The people in the CIA essentially know this, but it's not clear that our policy is as precisely directed as it might be.

QUESTION: The Colorado experiments that showed increasing "extremization" of views and convergence of views could have two underlying causes. They could be a real change in people's opinions or they could be a crumbling of inhibitions about expressing those. If the latter, you would be less worried about it, I think.

Secondly, those changes—the willingness to speak in more extreme and more convergent ways—might be very long-lived or very short-lived, and if the latter, you would be less worried about it.

Are there any experiments, or do you have ways of approaching those issues?

CASS SUNSTEIN: Those are also great questions.

We know enough to know that sometimes what happens is that people have a view, but they kind of suppress it, and once they are around likeminded people, it comes out. Then they are firm on it. What used to be described as consciousness-raising in the feminist community—that is that, usually. Where there were suppressed grievances that emerged, the inhibitions crumbled.

The only study I know on this isn't a controlled experiment, but it's the next best thing, which is a very careful account of what has happened to the disability movement. It shows that among the deaf community you have inhibitions about identity and about pride and about possibility that are relieved once people who are deaf are with one another.

On the other hand, we also have—and this isn't a controlled experiment, but this is also the next best thing—studies of ethnification, the best one involving the former Yugoslavia, where people who were marrying across ethnic lines, who were having dinner across ethnic lines started hating each other, as a result of something like the Colorado experiment getting going and then accelerating. In a couple of years, not long-simmering hatreds but created hostilities, historical incidents that people didn't even know about, became symbolic of longstanding grievance, as if human beings lived for 300 years and they were the same as their ancestors, whose grievance they didn't know about. So that wasn't a simmering one.

My suspicion is, in current Iraq—this is from someone who doesn't do this stuff professionally, so this is really amateur speculation— that it's mostly not like feminism but more like the former Yugoslavia, where this kind of intense self-identification is constructed rather than previously there but suppressed.

There are lots of theoretical papers on this. There are empirical studies of actual practices; no controlled experiments.

On short-lived versus long, there are a couple of things to say. The closest thing to an experiment there is Bennington College, where careful studies were done of what happened to Bennington College undergrads when they were there and how long the changes lasted. People went far to the left in Bennington College, in a certain period—I think the 1960s, maybe the 1950s—and it stuck.

We would expect, I think, that a little experiment—if you got a group of Americans together to talk about some fairly obscure issue, they might shift that day, but there is every reason, as your question suggests, to wonder whether they would think the same thing in a year or whether other information might change or they might think, "What was I doing on that day?" That is an important point.

Keep in mind, though, that you can think of the Internet as an iterated version of the Colorado experiment, in the sense that if people want, they just don't do this once and then live their lives; they enter Boulder or Colorado Springs a lot. So much of the intensity of political passion now is that people are living in information cocoons, and they don't exit them enough. It's as though they are always at Bennington or Brigham Young.

QUESTION: In New York there is a new program that started—I think it's probably entering its third year—called IQ Squared, which is a program that is designed to reactivate the old Oxonian idea of debate, for the question or against the question. There are always two people for the question and two people against the question. The questions are positive on issues such as environmental change. It was sold out practically before it started. It has really gotten rave reviews.

I wonder if something like that couldn't be done on the Internet very successfully, so that it would have a wider audience.

CASS SUNSTEIN: That's a great idea. The people who are involved in that maybe could put it on the Internet live and advertise this and invite questions from YouTube.

QUESTION: You sort of scared me when you started about judges and how you could determine how they would judge. A corporation that wants a certain answer just shops around for a judge. Is that justice in this country?

CASS SUNSTEIN: It's a little better and a little worse. To say this, I want to scare you a little bit more first.

The political party of the appointing president is a pretty good predictor of how judges are going to vote in ideologically contested cases. We now, with our 30,000 votes and our exhausted research assistants, have found that slightly scary, but not shocking fact.

Here is something that I think is slightly scarier, and it surprised us. If you want to know how a judge will vote, an even better predictor of how a judge will vote than the political party of the president who appointed him or her is the political party of the president who appointed the two other judges on the panel. So if you want to know how a Reagan appointee will vote, a really good predictor is, is that judge sitting with two Bush appointees or two Clinton appointees?

What makes it a little less worrisome from the standpoint of justice is that there is random assignment of judges to panels, and you often don't know until quite late. Sometimes you don't even know until the day of the argument. So a corporation can't shop for Reagan-Bush-Bush judges and a disability plaintiff can't shop for Carter-Clinton-Clinton ones.

But there is injustice, in the sense that similarly situated people, we know, are treated very differently, depending on the random draw.

We also can rank, and we have ranked, courts of appeals in ideological terms. We know which ones disabled people have a better shot at, and we know where African-American plaintiffs complaining of race discrimination are in big trouble.

One little funny part about this data, incidentally. We had a long time trying to figure out what's going on here. In all of the courts of appeals, with one exception, the pattern I have described exists. It's the usual beautiful pattern, where the the liberal voting rate decreases as there are more Rs on the panel, and Rs' votes get more liberal as there are more Ds on the panel. Rs are always less liberal than Ds, but it's climbing as there are more Ds there. So Rs are most liberal in RDD, and Ds are most liberal in DDD; Ds are most conservative in DRR, and Rs are most conservative in RRR, if you are with me.

We find that everywhere except one court of appeals, the Sixth Circuit. In the Sixth Circuit, Rs are very conservative, Ds are very liberal, and it doesn't matter who else they are sitting with. So Ds are not more liberal in DDD than in DRR; Rs are not more conservative in RRR than in RDD.

We wondered, why is this? It turns out, on the Sixth Circuit, the Democratic appointees and Republican appointees hate each other. They don't listen.

QUESTION: Where is the Sixth Circuit?

CASS SUNSTEIN: Midwest, roughly. [It consists of Kentucky, Michigan, Ohio, and Tennessee.]

QUESTION: I was wondering, in your studies in Colorado, how large were the groups?

Also, did you take age as a consideration? It would seem to me that people who live longer might not be swayed as easily as people who have just started in life.

CASS SUNSTEIN: Five- or six-person panels.

You would think that in the older and younger you might see a disparity, though it's not clear which way it would go. It might be that older people would be firmer in their convictions, less swayable. It might be that younger people are more fierce and passionate and less willing to get along.

We have that information. I will check it. It's a really good question.

I can tell you that every dimension along which group polarization has been tested, every demographic dimension, shows that men are the same as women, whites the same as African-Americans, well-educated the same as little-educated. So the basic phenomenon seems a great regularity. There is no identifiable group of people who don't fall into this, except the Sixth Circuit judges.

I have to tell you a little bit more that does support what you say. Remember, I talked about this beautiful pattern that we found in every court of appeals in the aggregate data. We have it now in about 23 areas. There are actually two areas where we don't see it, two areas where Democratic appointees are more liberal than Republican appointees and they are not affected by who they are sitting with. This would have implications for the Internet's effects on some people.

The two areas are abortion and capital punishment. In those areas, Ds are impervious to whether they are sitting with Rs, and Rs are impervious to whether they are sitting with Ds. The obvious reason is that these are issues where the convictions are very firm.

A slightly less obvious reason is that if people think of people who differ with them as really different along a salient dimension, then they don't listen. So if you had in the IQ Squared, here are the Democrats, here are the Republicans, and then you had the Republicans sitting in one place, the Democrats sitting in another place, and you made it all salient, then people would probably end up thinking a more extreme version of what they thought before—most people—even if they heard both sides.

So confidence and clarity do decrease polarization. The question is whether older people are more confident and more clear. We don't have any data that supports that. Maybe on some issues, such as Social Security.

QUESTION: Apropos of your discussion of developing more deliberative fora, perhaps you can get me back in touch with something I have lost touch with. I seem to remember that about ten years ago, an organization developed a weeklong seminar, I think at the University of Texas, to discuss the federal budget. A random selection of people from various political positions were selected and they were prepped with material in advance. By the end of the week, they managed to agree on a federal budget—a wonderful example of how people can be brought together.

Is that experiment still going on?

CASS SUNSTEIN: Yes. If I'm not mistaken, this is this guy Fishkin, who was at Texas and is now at Stanford. He has done a series of experiments in this vein. They are getting a lot of international attention. There are some countries, including China— not the most likely place for a deliberative opinion poll, maybe—where the governments are paying attention to what happens when people get together and talk across ideological lines.

QUESTION: Have you ever found that perhaps the most effective criticism comes from within? In other words, a liberal criticizing a liberal position is going to be more effective than a conservative doing the case, and vice versa.

CASS SUNSTEIN: That's an excellent point. If Rudy Giuliani says nuclear power deserves serious attention, that's not going to move the country much, but if Hillary Clinton says, as she recently did, that nuclear power deserves serious attention, that's a big deal. If Senator McCain is strongly for aggressive action to control greenhouse gases, that will have more impact, yes.

That does have implications for how to structure these deliberative fora.

As you may have picked up on, an increasing interest of mine is this idea of political charity. There is a connection between that and what you are saying. If leaders who are self-identified as Republicans are willing once in a while to say, "Democrats are right on this"—or not that, but say, "The following position X is right on that"—and Democrats do the same, that has an impact.

JOANNE MYERS: Cass, thank you so much for opening our eyes to this.