A post written a week ago about the popularity of Flickr users features effects of free versus pro accounts and the “Flickr Star System.” Now to further stimulate interest on Flickr, we can discuss a model of Flickr’s social cascades, proposed by Meeyoung Cha, Alan Mislove, Ben Adams and Krishna P. Gummadi in their article “Characterizing Social Cascades in Flickr,” which can be found at: http://www.mpi-sws.org/~amislove/publications/Cascades-WOSN.pdf

In this paper, the rate at which a picture is disseminated through Flickr users’ favorites is modeled after a formula for calculating R­₀, the epidemic threshold, which indexes the rate of disease spread through infections. R­₀ = 1 is the critical point indicating characteristics for epidemic outbreaks. A threshold of less than 1 indicates that a small number of initial infections will eventually die down from creating more infectees; a threshold of greater than 1 indicates epidemic growth. Naturally, public health aims to keep the threshold below 1 to stop epidemics. The adaptation of this model to social cascades in Flickr concludes that a popular photo’s reproduction rate is much higher than that of an infectious disease. But furthermore, this investigation produces more interesting characteristics of cascading in Flickr by examining the rate of photo popularity cascades.

While considering a Flickr user’s duration of exposure to a picture before he marks it down as a favorite, the authors of the paper also proposed a point which enlightens the class discussion of the roles of the few but significant individuals in cascades. These individuals, as the previous Flickr post points out, may be considered as the small number of (considerably the more popular) users with Pro accounts. However, the paper suggests that less popular individuals in Flickr, in a way, can be more “influential” in the cascades of popular photos in the long run. We must take into account the bookmark feature of favorite photos. As the number of such marked photos increases, older photos are less exposed to other users than do the newer favorites.  So while the popular users have more social contacts and therefore are more likely to disseminate more photos faster, they also replace their favorite photos at greater speeds. So if we are considering the dissemination of a particular photo, then a less popular user, who keeps the particular photo longer on the top of his favorite list, in fact increases the exposure time of the picture.  Hence, while significant individuals may spread more content to more people, the less popular individuals can actually be the ones responsible for spreading single particular items to more people in the long run. This interesting point emphasizes the significant differences that should be considered for different types and situations of social cascades.