WEST DES MOINES, IOWA -- It was dial-a-president night at the Holiday Inn here. On one side of the sliding partition in the main ballroom, a boisterous group of businessmen laughed at a comedian and inaugurated new officers. On the other side, some 85 earnest Iowans participated in an exercise in instant democracy that was as fascinating as it was frightening.

As the seven Democratic presidential hopefuls answered questions in a Houston auditorium, the Iowans -- likely participants in next February's first-in-the-nation caucus -- watched a big-screen television picture of the candidates. Each of the spectators held a hand-sized dial, numbered from one to seven, connected by wire to a nearby computer.

As the candidates came on the screen, one after another, delivering their responses to insinuating questions from William F. Buckley Jr. and obsequious questions from former Democratic National Chairman Robert S. Strauss, each Iowan rotated the pointer in his palm to indicate the degree of comfort or discomfort he felt with what he was seeing. ''If you very much like what you're hearing, go to No. 7,'' Chris Wheeler, whose Seattle company owns the system, had told them. ''If you strongly dislike it, go the other way to No. 1. No. 4 is neutral.''

The Iowans had been recruited by the Washington polling firm of Harrison Hickman and Paul Maslin, screened to be sure they were active Democrats. They were paid $25 for the evening's work, which most of them clearly took seriously.

Out of sight, their individual responses were merged by the computer every three seconds and plotted as points on a line on a graph, overlaid on the telecast of the debate. On the backstage monitor, where a half-dozen reporters were watching, the waverings of the line measured the response of the 85-member jury seated just a few feet away.

It was instant, summary judgment -- as final as the thumbs up-thumbs down of the Roman emperors viewing the gladiatorial combats.

The sophistication of the Wheeler Associates system is breathtaking. Our monitor showed not just the average rating each candidate was getting at that moment, but the range of reactions -- the percentage of 1's, 7's and in-between scores -- telling us whether his comments were polarizing the audience or building a consensus.

Down in a corner of the screen, separate ''cells'' showed the scores being reported at that moment by women and men, young and old, liberals and conservatives -- 24 subcategories, in all, within the jury.

As one who has struggled with only intermittent success since 1960 to gauge or guess the likely public reaction to political debates, I was awed by the flood of information this system supplied. Like a prospector striking gold, I felt like shouting, ''Data!, Charlie, fields of data!''

Clearly, anyone assigned to do instant analysis of future political debates would find such a system a boon. But what does this kind of technology imply for our politics and government?

Wheeler argues that it is a perfect tool of democracy, providing privacy for the individual, freedom from coercive group pressure, and a computer guarantee that each person's opinion will be given the same weight: true one-person, one-vote democracy. It also allows people to register their reactions nonverbally, he points out, ending the requirement of literacy and articulateness which most other polling or survey techniques require.

That sounds wonderful, until you ask whether democracy is really separable from literacy and articulated thought. The Founders assumed that both the leaders and the citizens of this republic would exercise judgment on its laws and policies. They believed that only through such a dialogue would the national interest emerge. They worried about sudden surges of sentiment; yet in this system, there is literally no time for reflection before the reaction is measured and recorded.

The other concern must be what will happen to this information, now that it is available. This was a trial run -- and a successful one. My guess is that by next year, each of the networks and major newspapers will have contracted with Wheeler or one of his competitors and set up its own ''jury'' of citizens to react instantly to campaign debates, to major presidential speeches, and any other major pre-scheduled event.

The verdict will be available as soon as the event is off the air -- if indeed the event is not interrupted to update a round-by-round scorecard. And that verdict will inevitably preempt discussion of what was actually said.

Clearly, too, the candidates will use the data to refine their message ''to make it more effective,'' if not more sincere.

It's amazing technology. But it represents one more step from a representative republic to a direct plebiscitary democracy. The first depends on responsible officials reconciling articulated views of their constituencies; the second on sophisticated political manipulation of mass emotions and the magnification, through instant media feedback, of momentary sentiments. That's what makes it scary.