Skip to main content



Information Cascades in Flickr

Information Cascades (Chapter 16) occur when people observe the actions of others and then make the same choice as others have made independent of their own information signals. This occurs mainly due to two reasons: the fact other peoples’ behavior can convey information about what they know and observing this behavior can be rational sometimes. Secondly, the fact that there is some kind of direct benefit when you follow someone else. Information cascades occur in various real life situations like politics, marketing, social networks etc. But how fast do these cascades actually help disseminate information to the crowds?

In this paper, researchers bring out an interesting point that sometimes information cascades do not occur as quickly as one might expect. They take the popular social network, Flickr and analyze the factors due to which information cascades are delayed. Particularly they investigate two factors that might have caused significant delay:

1. Bursty login times: Because users need to login to browse photos they have a window of time (login session) to receive content from their neighbors. Therefore, this factor directly contributes to the delay in information propagation

2. Content aging: As new photos are uploaded the availability of a photo from a particular user diminishes. As a result, additional clicks are required to access old photos which further delays the ability of photos (in this case old photos) to spread.

These two factors impose a crucial constraint on information propagation as the user has to not only be active but the content from his friend circle should be available for him during that time period. As a result of this, information takes a long time to propagate through the social link which means that popularity of content usually remains steady over time. Users are unable to imitate other’s behavior as they rarely get input. On a higher level this means that it usually takes long time to obtain an aggregate population outcome in Flickr.

What does this mean to us? This is a good example of the fact that cascades can be fragile and based on very little information. Since users are only logged in for a window of time which limits input from others their decision can be often based on very little information that gets passed on. Similarly as new photos get uploaded cascades from previous photos stop which creates a further delay. Therefore, although information cascades in various other markets such as the buying market have a tendency to spread quickly, in Flickr and other social networking websites with similar behavior they may actually experience a delay.

http://www.sciencedirect.com/science/article/pii/S1389128611003951

Comments

Leave a Reply

Blogging Calendar

November 2013
M T W T F S S
 123
45678910
11121314151617
18192021222324
252627282930  

Archives