SAN JOSE, CALIF. — No one in attendance here at the third International AAAI Conference on Weblogs and Social Media needed convincing that social networks contain a wealth of information. So Duncan Watts was quick to admit that his presentation yesterday morning, to a room full of academics who focus on the dynamics of social media, was like "bringing snow to the Eskimos." What Watts, a principal research scientist at Yahoo! Research, brought into the discussion was how these Web-based networks are revolutionizing the social sciences.

[Editors' Note: For more on the International AAAI Conference on Weblogs and Social Media, organized by the Association for the Advancement of Artificial Intelligence, see our blog for daily collections of Jessica Tsai's live-twittering -- Day One and Day Two. Her Day One coverage includes several links specifically related to Duncan Watts.]

Watts compared the rapid change in the way businesses are reaching their consumers to the Cambrian explosion — an evolutionary phenomenon that took place on Earth over 500 million years ago with the sudden burst of complex creatures. In a parallel sense, he told the crowd, a social explosion is what's taking place on the Web, leading to the development of different business models and new ways of advertising. He added, however, that while online relationships are quickly becoming a social norm, their "rules" are far from established. Relationships we've learned to resolve offline, he says, have yet to be "solved" online.

Even so, dynamic online relationships are influenced by -- and to a large extent reflective of -- these offline social behaviors. In this sense, Watts said, the Web is a "tool" for observing individual-level behavior and interactions between individuals in a larger population. Ever since sociology and economics have been established disciplines, he noted, sociologists have been aware that group interactions were critical to understanding the individual: Put multiple individuals together and they inevitably become a singular entity -- a family, a culture, a market. The problem, he said, is that until social networking came along just a decade ago "we haven't been able to measure these interactions except on a small scale, in a small environment, with crude tools that were snapshots based on survey responses."  

Because group dynamics were difficult to measure, much of the research that exists today bypassed this fundamental element, opting instead to focus on individual behavior. Watts referred to sociologist Allen Barton [LinkedIn profile here], who, according to a citation in the 2003 book Diffusion of Innovations, by Everett M. Rogers, in 1968 characterized the prior 30 years of empirical social research as having been dominated by the sample survey. Barton's view was that surveys' reliance on "a random sampling of individuals" -- which, as it turned out, continued to underpin sociology precepts for another three decades -- was little more than "a sociological meat-grinder, tearing the individual from his social context and guaranteeing that nobody in the study interacts with anyone else in it." The result? As Barton also noted, sociologists relying on surveys to assess a social organization may glean insight into "every hundredth cell…[but] structure and function are lost." In other words, without an understanding of interaction, it's impossible to truly know how the individual units are operating.

In his presentation here yesterday, Watts highlighted several instances of sociological research that he claimed would not have been possible prior to the social Web:

• Using email exchange to track social networks evolving in time: By measuring nine months' worth of intermittent transmissions between 40,000 individuals (students, professors, staff, alumni, etc.) at Columbia University, Watts studied the evolution and stability of social ties. New ties were most likely to form between individuals who shared a mutual friend and were part of the same group. To be just one link further apart, individuals were 50 times less likely to become friends. In terms of stability, Watts found that, at the aggregate level, relationships are stable over time, but high turnover exists at the individual level. Only about half of those who were highly connected at the beginning of the testing period were equally connected by the end.
• Using a Web-based experiment (Music Lab) to study the collective consequences of social influence on decision-making: Watts studied the phenomenon of how a "market" determines what will be a "cultural hit." His team created a fake environment called Music Lab where 48 songs by unknown singers were presented to the subject. If the person liked the song, they could download it. Individuals were heavily influenced by "Top 10" lists, which reaffirmed the notion that people rely on insight from others, but still does not explain what specific factors determine whether a book, movie, or song will be a hit. Watts suggested that this social phenomenon embodies "inherent unpredictability," which cannot be resolved regardless of the amount of subject data collected.
• Using a social networking site (Facebook) to study the difference between perceived and actual homogeneity of attitudes among friends: Contrary to arguments that Americans tend to gravitate toward cliques of like-minded individuals, Watts found that individuals are actually bad predictors of their friends' opinions. Based on data gathered from the Facebook application "Friend Sense," Watts reported that individuals are "terrible" at knowing when their friends disagree with them -- undermining the conventional wisdom that friends are strong influencers.
• Using Amazon's Mechanical Turk to study the incentives underlying crowdsourcing: Watts found that, regardless of how people are compensated, they always feel underpaid. Individuals base their expectations off an "anchor," or one piece of information -- hence the influence of a cognitive bias known as the "Anchoring Effect."

News relevant to the customer relationship management industry is posted several times a day on destinationCRM.com, in addition to the news section Insight that appears every month in the pages of CRM magazine. You may leave a public comment regarding this article by clicking on "Comments" at the top; to contact the editors, please email editor@destinationCRM.com.