In Social Networking, 'Weak' Connections May Be the Most Powerful

One of the upsides to the otherwise mixed blessing of online social interaction is the production of a written record that might be studied by social scientists. So, entire models of social interaction might be rewritten according to how information spreads on Facebook.

For instance, Emilio Ferrara, a data science and machine learning professor at Indiana University, and a team of Italian researchers just released a paper that revisits some of the foundational concepts behind how social networks actually work. As those concepts were developed by Stanford sociologist Mark Granovetter back in the 1970s, we're talking about social networks in the old sense of the words.

Granovetter's work on strong and weak tieshas been cited almost 30,000 times, according to Google Scholar, so revisiting it with current data is an ambitious undertaking. But, according to Ferrara, it's also a necessary one.

The team behind the current study started by updating the definitions of what qualifies as a strong or a weak tie to more accurately describe what they observe in real life, or more specifically, real life as it happens on Facebook.

"In the original definition of Granovetter this is a very complex task, because you're supposed to know a lot about the interactions of individuals–the high-resolution," Ferrara told me. "So, how often they see each other; how much time they spend together; how often they communicate or interact; [how often] they talk they chat they see each other and so on."

"In our definition—which is a little bit more simple, but also much more effective for nowadays' society—we just look at how these people are connected via friendship," Ferrara said.

Thus, "strong ties" become ones that are within communities, people with whom a user shares a lot of connections and friends. "Weak ties" then are those between members of different communities, with fewer shared connections.

The biggest difference between what Ferrara and his team found using their definition versus the prevailing wisdom is that people actually have a lot more weak ties than strong.

"There is something really kind of fuzzy at this point in social network theory," Ferrara said. "So there's a very sharp contrast between what social theories will tell us and what we observe in the real world, when we look at real large-scale social networks like Facebook. It's something we're not understanding very well or very clearly and there's sort of a disconnect between our theory and observations."

"So the fact that the theory would tell you, to some extent, [is] that what we are observing online is not possible or is unlikely to happen," Ferrara added. "It would tell you that we don't have that many weak ties, but what we observe is that as much as about 80 percent of ties are weak."

But that's not a value judgement. The consequences of severing weak ties can actually be more severe than severing strong ones. It's easy to understand why: If someone loses a connection with a contact that they share a lot of friends with, whatever information they get and pass on can be acquired via other channels, through other friends. Losing a weak tie, however, cuts off a channel to another community and, thus, access to what that community offers—be it a new job or just information.

"It's really hard to remove the connectivity of a cluster or community from another, there will still be connections," Ferrara said. "Of course even when you remove these interconnections, information still flows, but in a much more inefficient way. Much more slowly."

Knowing how information spreads, where it creates dark pockets isn't just interesting to marketers trying to make sure the world knows when the McRib is back, although there's definitely something for them here. Imagine reaching people in the path of a storm, or in the midst of an epidemic, with relevant information. The old model is ripe for reinvention. Tell everyone you know.