My friend Tom Corddry recommended Connected: The Surprising Power of Our Social Networks and How They Shape Our Lives. Since Tom knows more about this stuff than anyone else I know, I expected great things, and I was not disappointed. Note that this is a book about social networks, not social media. While much of the information can be applied to on-line communities, the book is not about them in isolation. If you are looking for something about Facebook, keep going.
This fascinating book provides a lot of insight into phenomena that are otherwise difficult to understand. For instance, the authors demonstrate that the motivation to vote derives from social networks. From a purely rational, economic perspective, it does not make any sense to vote. Let’s say you would pay $1,000 to be the only person who chooses the winner in an election. An economist would say that by voting you are buying a lottery ticket with a potential payoff to you of $1,000. You “win” the lottery only if there is a tie between the candidates and your vote becomes the deciding vote. Guess how often that happens? Given situations like Florida 2000, you might think it happens from time to time. Well, it has never happened even once in any election, Federal, State or local, during the entire history of the United States. This is a lottery that you have no chance of winning. Economically, it is literally not worth the gas to drive to the polling place. So why do people vote? The authors suggest that citizens know something instinctively that economists do not. They know that by voting, they will influence others to vote as well. As a result, there does not have to be a tie in order to win the lottery. Voting is, to some degree, contagious. It turns out that this is the case with most things, whether it is the urge to yawn, one’s emotional state, disease, weight gain or loss, attitudes, information or an idea.
The authors call the likelihood that the act of one person will influence others “amplification.” In the case of voting, a conservative voting might inspire a liberal friend to vote because the liberal friend might want to “balance” his conservative friend. Not surprisingly though, amplification works best in networks of relatively similar people. A single liberal voting will influence many more liberals to vote than conservatives. The distance and rate at which things spread through a network are a function of the network’s structure — how “transitive” the network is. In networks with high transitivity, most of the members know most of the other members. In networks with low transitivity, most of the members only know a few of the other members, but all are still connected — just in a more linear way. You might think that things travel farther and faster through networks with high transitivity, but the authors state that is not the case. Highly transitive networks can have insular clusters where individual participants can’t influence individuals in other clusters of the network. People who are more moderately transitive are more likely to act as bridges between clusters in a network. In other words, transitivity needs to be just right. When you have the right mix of amplification and transitivity, the results can be dramatic. The authors talk of voting “cascades” where one voter influences hundreds and ultimately perhaps thousands of others to vote. This is possible because you can be influenced by people in your network that you don’t even know. You may have a friend A who in turn has a friend B whom you do not know. Friend A is apathetic about voting, but friend B votes and, as a result, friend A feels that the act of voting is more important than before. Friend A communicates this new attitude to you, and you run down to the polling place to vote because of the influence friend B exerted upon you.
The potential applications are many. Consider this recent Bloomberg story on how the SEC is deploying novel techniques to identify insider trading:
[T]he SEC began using computer software about two years ago to sift hundreds of millions of electronic trading records, known as blue sheets, attached to the stock exchange reports about suspicious incidents, according to people familiar with the project. By looking for patterns in the library of data, they identified groups of traders who repeatedly made similar well-timed bets.
Once investigators find a cluster of correlated trades, they tap other sources of information to unravel how its members obtain and share tips, the people said. For example, if a group profits on trades before a series of corporate takeovers, the SEC may check so-called league tables listing which investment banks or law firms advised the deals. If one firm was involved in all of them, an employee there may be the source of the leak.
Can you find more customers the way the SEC is finding inside traders — using their social networks to detect them? Even better, can you create a cascade of new customers voting for your product or service by finding more effective ways to influence them by understanding the structure of their social networks?