Indifferent Attachment: The Role of Degree in Ranking Friends

The MySpace social networking site allows each user to designate a small subset of her friends as “Top Friends,” and place them in a rank-ordered list that is displayed prominently on her profile. By examining a large set of \(\approx \)11 M MySpace users’ choices of their #1 (best) and #2 (second-best) friends from historical crawl data from when MySpace was more popular than it now is, we discover that MySpace users were nearly indifferent to the popularity of these two friends when choosing which to designate as their best friend. Depending on the precise metric of popularity we choose, the fraction of users who select the more popular of these two individuals as their best friend wavers above and below 50 %, and is always between 49.3 and 51.4 %: that is, the popularity of the two candidates is essentially uninformative about which will be chosen as the best friend. Comparisons of other pairs of ranks within the Top Friends (e.g., #1-versus-#3, #2-versus-#3, ...) also reveal no marked preference for a popular friend over a less popular one; in fact, there is some evidence that individuals tend to prefer less popular friends over more popular ones. To the extent that ranking decisions form a window into broader decisions about whom to befriend at all, these observations suggest that positing individuals’ tendency to attach to popular people—as in network-growth models like preferential attachment—may not suffice to explain the heavy-tailed degree distributions seen in real networks.