Facebook found the same effect as described in this article. People are friends with those who are similar to them.
Information is widely shared between friends. It is only because people have so many acquaintances that information is spread into clusters of close friends. However, that information flow from acquaintances isn't increasing the diversity of beliefs within a cluster of friends. If it was, then you would see clusters of friends with diverse beliefs which has previously been demonstrated to not be the case.
The example given in the Wired article about the two universities in Kansas implies that the students at the larger university, who are surrounded by a more diverse students but who have close friends similar to themselves, are exposed to less diverse opinions. That is never actually proven in the article.
https://www.facebook.com/notes/facebook-data-team/rethinking...