These comments were presented on February 28, 2009, at an American Academy of Arts and Sciences symposium on The Public Good: The Impact of Information Technology on Society. The symposium was co-sponsored by Microsoft, Google, and the Computer History Museum, and Cohen participated along with Henry Brady and Edward Felten in a panel on Information Technology and Democracy.

Democracy is a political system in which elections and other devices are intended to hold officials accountable to the people. But while elections and other institutions of accountability are essential to democracy, they are just part of the story. Their efficacy arguably depends on a complicated, dispersed, ongoing, relatively unstructured public discussion: communication in what Jürgen Habermas calls the “informal public sphere.” Elections focus the conversation, but public discussion enjoys a life outside the rhythms of formal politics: at work, in churches and parks, in families and at school.

How does the Internet shape this informal process of discussion—public discourse?

If you asked me to summarize everything that can be said with confidence on this subject, I would be done now: the technologies are changing too fast (Facebook and Twitter are too recent to have generated serious analysis of causal consequences, as distinct from illuminating descriptions of usage patterns), the phenomena are too amorphous, and the historical benchmarks for measuring improvement are too often lacking. Addressing this question is like asking an epidemiologist about the impact of a medical intervention on diseases of unknown etiology, in a migrant population with no medical records, living in an environment with a high rate of viral mutation. Fortunately, I am a political theorist (and a magazine editor), not an empiricist, so I will jump in with both feet.

I want to distinguish four dimensions of concern that should occupy our attention when we think about information technology and public discourse. To see the dimensions, start with a model of an ideally conducted discussion about public affairs, including public policies (health care, taxes, Iraq war), as well as the values and principles that ought to guide common affairs. The discussion would have four qualities: it would be deliberative (reasoned); it would ensure equal and low-cost access to information; it would provide equal opportunities for public influence, to present arguments to others; and it would be founded on information of high quality. My judgment, in a nutshell, is that the jury on deliberativeness is still out; that on opportunities for access to information and for public influence, things look pretty good, though with some concerns; and on high-quality information, we have serious reasons to worry because of the Internet’s impact on newspapers (in particular, on investigative journalism). Consider these points in turn.

First, deliberativeness: an interesting debate has emerged about whether the Internet generally and the blogosphere in particular fosters deliberation—mutual engagement and argument across differences of view—or instead invites people to cloister themselves by filtering information and opinions that do not ratify preconceptions. Are we getting lots of backslapping among the likeminded, which, in all sorts of unsurprising ways, encourages more extreme views? Or are we getting discussion among people who disagree, which—according to lots of evidence—gives us a better understanding of the views of others and of our own views and promotes a more tolerant attitude?

Focusing on the blogosphere, what are the basic facts about cloistering and polarization? First, that the likeminded talk much more to each other than they talk to people they disagree with: only 10–15 percent of links on a conservative blog will point to a liberal blog, and vice versa. But, second, it is not at all clear that this represents greater self-cloistering than we find elsewhere in informal discussion (as Diana Mutz and others have argued, the workplace is the big exception). Third, the reason for the self-cloistering is arguably that bloggers are a select community of strong partisans, not that blogging polarizes. And fourth, there is no evidence (not yet, anyway) of increasing self-cloistering and polarization. So on this first dimension, no reason to go out and celebrate, but not much to fear.

Second, on access to information, the story looks good. The essential point is that costs of acquiring all sorts of information relevant to making public judgments have markedly diminished. Three more specific points: first, I am emphasizing access to information and measuring access by costs. I am not saying that there is equal access (work by Henry Brady, Kay Schlozman, and Sid Verba describes the degree of inequality), but that the costs of access have diminished for all. Second, opportunity to acquire information is not the same as being informed, and there is no evidence for the proposition that the Internet treatment has produced, on average, better-informed citizens. Still, greater opportunity is important. Third, some analysts wonder about the reality of improved access to information because they observe that Web and blog traffic follow a power-law distribution, with a small number of sites drawing the lion’s share of the traffic—from which one might conclude that this world is not really so different from the pre-Web world of a few dominant, relatively inexpensive sources of information (radio, television, newsprint). We should resist that inference: the power-law distribution matters, in a way that I will describe in a moment, but does not defeat the point about greater access.

On the third area of concern, greater opportunity for public influence, the results seem not bad, but more mixed. The basic idea was nicely captured in an article that Mitch Kapor wrote for Wired and Boston Review in the early 1990s about the prospect of a Jeffersonian new information highway: “Ideally, any user of any network should be able to originate programming from the home. . . . it should be as easy to provide a service as it is to use one.” Here I want to refer to my own experience as a magazine editor: Boston Review went online in 1995, when Mosaic was the browser of choice. We now print 10,000 copies every two months, and have 200,000+ unique visitors online in the same period: we have vastly greater reach than would be possible with paper alone. Of course the point generalizes beyond the personal anecdote: we have not arrived at Kapor’s Jeffersonian world, where it is as easy to provide a service as to use one. But we are closer, as the tens of millions of blogs indicate. That said, when we are thinking about chances for influence, the power-law distribution is important: although one-many communication is much easier, most of us languish in the long tail, with a high probability of being swamped. Still, there is a large difference between a world in which the costs of one-many messaging are relatively small and in which the costs are prohibitive, even if the audience distribution looks like a power law.

Finally, quality and depth of information and analysis. Here, we face a large danger. I went to a talk last year by Jonathan Zittrain, author of the wonderful book on The Future of the Internet—and How To Stop It. At some point during the talk, Zittrain pointed to Craig Newmark of Craigslist who was in the audience, and said, to an enthusiastic reception: there’s the guy who brought down the American newspaper business. I was not sharing the good cheer. But attitudes aside, it is true that with income from classifieds cratering, huge difficulties getting people to pay for newspaper content, and limits on online advertising income, newspapers lack a business model. That is not good news. Simply stated, we cannot have a successfully democratic public sphere without the kinds of information that newspapers have supplied. I am not thinking of weather reports, but of investigative journalism, local, national, and international. Here, the Internet treatment has been damaging, the damage is growing, and the consequences are potentially severe.

Of course investigative journalism is not married to newsprint, but that is how it has been presented, and there is not yet a viable alternative: 15,000 journalists lost jobs last year, cuts at large newspaper are running at 20 percent. The blogosphere is no alternative in terms of investigative resources, access, legal-defense capacity, or training and mentoring. It lives parasitically off investigative journalism. Call me old-fashioned, but poetry, philosophy, physics, and investigative journalism cannot be blogged and crowd-sourced.

There is a growing debate about the best successor model to commercial newspapers, with many contending proposals. I will mention three, each of which assigns a large role to an electronic public sphere: (1) private foundations or donors either provide endowment for newspapers or for nonprofits that employ political journalists (Propublica is the leading example, with editors and twenty-eight journalists who provide content to print and online media); (2) a public system that would extend the public broadcasting model to print media; and (3) a national endowment for journalism, with support tied to audience size (proposed by Bruce Ackerman and Ian Ayres).

This is not the time or place to explore these alternatives. But in this fourth arena, cyber-utopianism—a celebration of the dispersed, decentralized, egalitarian, Jeffersonian, participatory, deliberative electronic public sphere—is not only misplaced but dangerous.