Net Election

The Next Killer App 

Slate, the Industry Standard, and join forces to examine the effect of the Internet on Campaign 2000. 

After a year of examining the role of the Internet in Campaign 2000, have we learned anything? To my mind, one of the most important lessons was underscored by all the hype over the transforming effects of 360-degree Web cams, streaming media, and the like. And the lesson is this: For an Internet technology to take root in the political arena, that technology must have already established itself somewhere else.

That may sound obvious, but it has been learned and relearned over the course of this campaign. E-mail, e-commerce, discussion lists—the list of political Internet success stories reads like a list of the Net’s killer apps. And soon, you’ll be able to add collaborative filtering to the register.

Collaborative filtering has been around for a while. The pioneer was a company called Firefly that developed a music-recommending program using collaborative filtering in 1995, the Mesozoic era as far as Internet time is concerned. The technique is now fairly common. It’s how knows what books to recommend to you. It’s how a site called MovieLens knows what movies you’ll like.

How does it work? Massachusetts Institute of Technology professor Pattie Maes calls collaborative filtering “electronic word of mouth.” It’s software that uses information about your tastes—in movies or books or music or whatever—to predict what else you might like. As Malcolm Gladwell put it in thisNew Yorker article last October, ” … [I]n order to know what someone wants, what you really need to know is what they’ve wanted.” It sounds obvious. But traditionally marketers have targeted demographic groups by age, sex, race, income, ZIP code, etc. Using collaborative filtering on the Internet, they’re increasingly targeting “preference groups” instead.

Right now, most politicians are still using the blunt instruments of demographic marketing. They market themselves in “swing states” or to groups like “soccer moms,” who are supposed to represent the elusive swing voter. But swing voters are really a preference group (multiple preference groups, actually), not a demographic group. Collaborative filtering has the potential to slice and dice the electorate into preference groups so that they can be targeted with more precise and effective messages. And with that potential comes familiar worries about electronic privacy.

If these concerns have been muted so far, it’s because politicians aren’t really using collaborative filtering yet. But at least one political site is already using collaborative filtering in a less controversial way—to improve political discourse on the Web. uses the technology to create order out of the chaos of “digital town halls.” If the Sunday talk shows are “Pundit Central,” online town halls are the Pundit Periphery. Many users come to the Net hoping to engage in lively, high-minded discussion only to tire quickly of rants, flame wars, and commercial spam. Without collaborative filtering, the only way to moderate the discussion is to use a top-down approach, like the tightly controlled method used at Al Gore’s town hall.  Even a slightly more open approach, such as the one used by California U.S. Rep. (and U.S. Senate candidate) Tom Campbell, limits the number of questions that can be discussed. uses collaborative filtering to moderate the discussion from the bottom up. Users fill out a short questionnaire about their political values and issue priorities. This is used to sort them into political preference groups such as “New Dealers,” “Pro-Business Progressives” (New Democrats), “Moral Conservatives,” “Economic Conservatives,” “Skeptical Independents,” and the like. Postings are then displayed by “relevance” to each user, based on the rating of the posting by other site users and the shared political values of others who have rated the article, among other things. This doesn’t mean that you read only the things that you agree with—it means that you read things that you are more likely to be interested in, which probably won’t include rants, flames, and spam. (Similar systems are used to discuss technology at and Jewish issues at

The site is also staging an e-debate between the senatorial candidates and congressional candidates in the Philadelphia area. Unlike Web White & Blue’s “Rolling Cyber Debate” or the televised “town hall” presidential debate, the voters’ questions were not pre-selected by a moderator, but by the group as a whole. Unfortunately, this doesn’t seem to make the questions any more interesting. Still, the event bodes well for the development of interactive political events. The debate includes not only the questions that were asked of the candidates, but also the questions that were not asked, and there is the potential for voters to discuss what the candidates said, what they didn’t say, and why.

One problem with the site is that there aren’t many users yet. A second problem is more vexing. In debates such as the one hosted by, there doesn’t seem to be a role for the rude but important question—the kind that some political journalists are good at asking. While the site is a welcome and much-needed addition to the political Web, empowering voters to interact directly with politicians won’t eliminate the need for journalists and experts, any more than MovieLens replaces movie critics. Or at least I hope it won’t.