a critical mass of easily influenced people

In an update on his recent research on simulated influence networks, Duncan Watts challenges a model of influence popularized by Malcolm Gladwell in The Tipping Point which tells us that the rapid, cascading social changes (fads, crime epidemics, etc.) are largely initiated by a small number of highly influential folks with attributes that make them exceptional social connectors, persuaders, and knowledge brokers. Watts and his partner Peter Dodds ran thousands of simulations of behavioral change across networks of agents, testing a range of variables related to influence. They found that, in most cases, "global cascades" of behavioral change did not rely on the characteristics of the small number of influentials populating their models, but rather on "the availability and connectedness of easily influenced people."
In other words the propagation of influence depends more on the structure of relationships between "ordinary folks" than on the special characteristics of an expert or charismatic few.
One the one hand, this is encouraging. "If the network permits global cascades because it has the right concentration and configuration of adopters, virtually anyone can start one." Apparently, there's hope for the little guy after all, as long as there are a great enough number of other little guys nearby, paying enough attention.
On the other hand, Watts' article dismisses the kind of influence that celebrities like Oprah Winfrey wield as an "exception..a function of media, not interpersonal influence." Sure, new media, Web 2.0, etc. are initiating enormous changes in the cultural environment, but they enter an established environment structured on broadcasting principles that amplify the power of a small number of voices, and these voices are as constant a presence in many people's daily lives as those of their neighbors, family, and friends. New media is amazing, but most people still get their news from TV.
The final question is: what is a potential influencer supposed to do with Watts' insight? If your thing is selling soda or electing politicians, it so much easier to invent an inventory of attributes that define an "influential" and then reach out to folks with those attributes. But if what you really need is a "critical mass of easily influenced people" where do you get some and what do you do with them? At the close of the article, Watts briskly invokes "Web-based social networking tools" (read: Web 2.0), but the phrase "critical mass of easily influenced people" makes me think of television, not YouTube or MySpace.