We have just completed another round in a continuing national experiment in political theory—the primary selection process as it has been revised in several waves of democratic reform. I believe this experiment, filled with noble intentions, has largely been a failure. From the standpoint of democratic theory, the presidential selection process should be both representative and deliberative. The process should be representative of the country in all its diversity and deliberative in permitting an informed consideration of the issues and the candidates. At the moment, the process must be judged inadequate on both criteria.

The process fails to be deliberative because reforms have brought power to the people while, at the same time, the people have little incentive or opportunity to think about the power they exercise. Citizens have been rendered prisoners of soundbite democracy by what social scientists call a “collective action” problem. While more citizen deliberation about the public good would likely serve the collective welfare, individual citizens have so little effect on the outcome and so few opportunities for political efficacy that it is easy for them to tune out, to feel “disconnected,” as a recent Kettering Foundation report documented. While there is, theoretically, a great deal of information about the election and its issues available from published materials, from C-Span, even from books written by candidates, citizens have little incentive or effective motivation to invest their time and effort in that information or to deliberate about it. And though there has been a great deal of experimentation with proposals aimed at improving media coverage of this year’s campaign, even a cursory review of these experiments makes it clear that something more dramatic is needed if we arc to involve the citizenry in a process that is both representative and deliberative.

Since 1968, when Hubert Humphrey won the Democratic nomination without entering a single primary, a series of well known reforms has led to the proliferation of primaries. The McGovern-Fraser Commission announced that “the cure for the ills of democracy was more democracy.” In moving from 17 to 37 primaries from 1968 to 1988 (and 39 this time), we have created a system for both major parties that is neither deliberative nor representative. “Front-loading” has made the assessment of most candidacies extremely superficial. The insatiable need for money to reach large media markets leads candidates to drop out if they fare poorly in the initial winnowing. The voters in Illinois and Michigan, much less Pennsylvania, never got to pass on Harkin, a candidate who had natural constituencies there. Most importantly, Clinton won the Democratic nomination because he won the “invisible primary”—the quest for credibility, fundraising, and organization before the official primaries begin. Tsongas said as much when he dropped out.

Other irrationalities of the current primary process deserve at least brief mention. Crucial contests turn on self-selected primary electorates, unrepresentative of their states in the general election and disproportionate in their influence on the process because of their effects on momentum with the next contest. Fifty percent of the voters in the Maryland primary were at the college-graduate or post-graduate level in education. Blue-collar voters simply did not show up in that primary. The Florida primary was dominated by voters over age 60 and by Jewish voters, groups that were both susceptible to Clinton’s negative ads about Tsongas on Social Security and on Israel.

Whatever the merits of New Hampshire as a starting point, we now expect masses of voters around the country to pass on candidates about whom they have only the vaguest impressions. As the process approached Super Tuesday, a New York Times poll of likely primary voters nationwide reported “no opinion” levels for the five leading Democratic candidates ranging from 52 to 75 percent. Note these are not levels reflecting that the voters are undecided, but whether they had any opinion at all about these candidates. Furthermore, to the extent that voters acquired an opinion, it was likely to be formed from impressions of smears in supermarket tabloids or impressions that a candidate has the image of being substantive, without the voters knowing very much about the issues at that stage.

Iowa and New Hampshire are, of course, unrepresentative of the country—ethnically, racially, politically, on almost any dimension you choose. The accident that New Hampshire is in dire straits economically conditioned the entire beginning of this campaign. Imagine if Harkin had not run and if relatively prosperous Iowa had started things off how different the initial debate and the framing of issues would have been. Remember that in 1988, Bush was bounced around by an economically depressed Iowa but restored by what was then a prosperous New Hampshire. The arbitrariness of the ordering of these events and the conditions in any one state make it imperative that we have a different way of starting things off. One study concluded that in 1988 media coverage of Iowa and New Hampshire voting equaled coverage of all the other primaries and caucuses put together. Political scientist William C. Adams has calculated that New Hampshire typically gets 105 times the coverage per voter as the large Ohio primary later in the process.

As sociologist Kiku Adatto demonstrated in a much publicized study, from 1968 to 1988, the average soundbite for presidential candidates—the period during which a candidate could speak uninterrupted on the network evening news— shrank from about 42 seconds to a little more than nine seconds. The amount of political discourse reaching the electorate is being reduced to messages worthy of fortune cookies or bumper stickers. If this election was true to form, all the minutes of substance on any of the networks’ evening news broadcasts from January to June will add up to less than two hours. By contrast, the warm-up to the Super Bowl this year was two-and-a-half hours.

While some published reports claim that the soundbites increased in length this year, such claims depend on including PBS and CNN in calculations for 1992, but not for 1988. Political scientists Marion Just and Ann Creigler (of the Joan Shorenstein Barone Center at Harvard) have recently completed a study of the crucial initial campaign period, from February 1 to 18, and conclude that for the three major networks, the average soundbite is not significantly greater than in 1988. Furthermore, the average soundbite for Brown was down to six seconds; for Tsongas, Kerrey, and Harkin, about nine seconds; and for Clinton, about 12 seconds. Clearly, during the crucial initial period of the campaign season, the networks were doing some initial winnowing of the visibility of the candidates. The Just/Creigler study of March shows deterioration even from these levels, in that Clinton’s soundbites shrank during the climactic period of the primary season to about six-and-a-half seconds.

The mass public has little opportunity or incentive to become informed and little incentive to deliberate on the issues. Our democratic mechanisms are really operating as what political scientists from V. O. Key to Giovanni Sartori have called an “echo chamber,” bouncing back surface impressions from television via polls and primaries. We need more than impression management; we need deliberation on the issues that somehow also represents the entire country.

The move toward mass, direct democracy in the large nation- state derives much of its appeal from an image of direct democracy reminiscent of the Athenian Assembly, or of the New England town meeting. But such an appeal is mistaken. The social conditions for face-to-face interaction and deliberation present on a small scale are not present in the larger nation-state. In primaries, referendums, opinion polls, and “teledemocracy” (such as the “electronic town halls” with viewer call-ins advocated by Ross Perot), we get the isolated, atomized citizen, pulling a lever, casting a ballot, or dialing an 800 number based on very little reflection or interaction. His or her vote is just one of millions that will have little effect on the outcome. The citizen has little incentive for informed debate or for investment in political knowledge. Even the most optimistic account of contemporary voter rationality, Samuel Popkin’s recent The Reasoning Voter, characterizes voter thinking in terms of “low information rationality”: the voter’s motivations can be generalized as “what have you done for me lately?”

The old source of deliberation was in the elite processes of the selection system. The dilemma we are concerned with is whether we have politically equal but relatively incompetent masses (in primaries, polls, and in referendums) or politically unequal but relatively more competent elites (in smoke-filled rooms). Neither is an adequate form of democracy.

The media responded to this dilemma by broadcasting an impressive lineup of nationally televised debates among the Democratic candidates. During the period when there were several active candidates, there were 11 nationally televised debates between December 15 and March 15. Seven of these were broadcast by local stations and C-Span. The C-Span national ratings were very small, but some of the debates relevant to particular primaries achieved modestly respectable local ratings. The other four debates were broadcast nationally by NBC, ABC, CNN, and PBS. Ratings of the major network broadcasts ranged from a low of 2.1 for the climactic CNN/League of Women Voters debate before the New Hampshire primary to a high of 5.5 for the ABC debate on March 5 during its Nightline slot (each rating point is a percent of the 92.1 million television households and represents 921,000 households). These ratings put the debates squarely in the bottom ninth of network programming in their respective weeks. Unlike the final presidential debates of the general elections, primary debates (when candidate selection is a live issue) have not attracted large audiences (the 1988 presidential debates in the general election were estimated to have drawn more than 160 million viewers).

My point in mentioning the ratings of these debates is to emphasize that the effective discourse reaching the public remains the shrinking soundbite on the network or local news of a staged photo opportunity when a candidate passes through immediately before the relevant primary. Furthermore, it is worth noting that these debates, when they were turned into soundbites and newspaper stories, were reported mostly in terms of whatever conflict, controversy, or confrontation they happened to generate. The first debate, on NBC, was most notable for the flap over Jerry Brown’s advertising his 800 number on the air and for Harkin holding up a dollar bill to symbolize the value of the middle-class tax cut. The CNN debate in New Hampshire was reported mostly in terms of the fire Tsongas drew for his support of nuclear power. The Denver debate is remembered for Tsongas’ response to Clinton that while he might not be “perfect,” at least he is “honest.” The Dallas debate was notable for Clinton’s rejoinder to Brown that he should “chill out,” a phrase that Hillary later took credit for in the press. The WLS-Chicago debate the Sunday before Illinois and Michigan was sparked by Brown’s exaggerated description of a Washington Post article about alleged conflicts of interest arising from Hillary Clinton’s law firm in Arkansas. These debates, with the possible exception of the MacNeill-Lehrer debate on PBS, were less than enlightening, even for those few citizens who watched them. Yet the principal difficulty is that to the extent that the debates reached the public, they did so primarily in terms of soundbites chosen for drama or conflict, soundbites that could be recycled on national or local newscasts. The debates, while a noble effort, did not improve the quality of the discourse reaching the public.

Another effort worth mentioning is the Discovery Channel’s offer to all the major candidates to speak directly to the public. Without the filters of pundits or editors, the candidates could reach the public directly for 20 minutes each. Unfortunately, the broadcast achieved a rating of only about 1.5, reaching about 1.2 million of the nation’s television households. Because the format was not conducive to drama or conflict, it was not widely reported and produced very few soundbites.

Imagine a new beginning to our season of presidential selection. Suppose we took a national random sample of the voting-age population and transported them to a single site. We invite the major presidential candidates for several days of face-to-face questioning in small group sessions. We provide the citizens with briefing materials beforehand on the major issues facing the country. At the end of these deliberations, we poll the citizens on their views of both the candidates and the issues.

Such an event would constitute what I call a “deliberative opinion poll.” In Democracy and Deliberation, I try to make the ease for deliberative polls as a response to the problems in our presidential selection system. An ordinary poll models what the public thinks, given how little it knows and how little it pays attention. A deliberative poll models what the public would think if it had a more adequate chance to think about the issues. The point of an ordinary poll is descriptive. It provides a snapshot of the public’s unreflective preferences. The point of a deliberative poll is prescriptive. It gives voice to the people under special conditions, whereupon the people would have a voice worth listening to.

I proposed to demonstrate this notion on national television (on PBS) in 1992. After the event was announced by WETA (the Washington PBS station), full funding was not achieved, and the event was unfortunately canceled for this presidential season. However, the event has now been adopted not only by WETA, but also by all ten of the nation’s presidential libraries, and we propose to start early for 1996. Note that a successful demonstration of the idea could transform the presidential selection system by turning the “invisible primary” into a deliberative event. Given the role of momentum in the invisible primary, a change in the way candidacies are launched could determine the results. Instead of an event like the Florida straw poll, which is both unrepresentative of the country and undeliberative, we would have an event that is both representative of the entire country in all its diversity and deliberative on the issues.

The Florida straw poll, which represents party activists in one Southern state, played a major role this year in Clinton’s rise to the front, just as it played a major role in Jimmy Carter’s emergence before the Iowa caucuses in 1976. Instead of candidates repeating standard stump speeches, we might imagine a process whereby they arc forced to respond in depth on the issues, with sustained follow-ups. The premium that applies to unrehearsed political discourse was dramatized by the incident last December when President Bush, in a teleconference to a teachers’ convention in California, complained that he was asked the questions in the wrong order. An apparently spontaneous event was actually one that was carefully scripted.

New innovations are necessary if we are to create a public voice for the expression of the will of the people. Deliberative opinion polls conducted at the beginning of the selection process could have a major effect on all that follows. It would represent a constructive use of the same three factors that have facilitated the rise of plebiscitary democracy—the impulse to bring the people into the process, the rise of television, and the development of public-opinion polling.