The study, Measuring online social bubbles [peerj.com], peered into anonymous data from 100,000 users over 3 and a half years -- a dataset that includes 18 million clicks on AOL search, and 1.3 billion tweets shared by over 89 million people.
Their data demonstrates [treehugger.com] that accessing information via social media [mnn.com] exposed people to more of the same sources and fewer overall sources. Although most of the data demonstrates only a collective social bubble effect, the researchers used a couple of smaller data sets in which the clicks could be associated to individuals to suggest that the behavior found in the collective analysis parallels the behavior in the smaller sets quite well -- the individual is most likely experiencing the bubble effect that the larger dataset portrays.
...
There is no fighting the current of a trend as strong as the online network, but perhaps we need to introduce a new spin. Could social media introduce something like the "I'm feeling lucky" button on google? But in fact, things will probably continue deeper into the narrows, as the web gets smarter about what we clicked in the past and tries to deliver more of the same.
In other words, living in a bubble is comfortable, but ill-prepares you for understanding the real world.