We find that ideological segregation of online news consumption is low in absolute terms, higher than the segregation of most offline news consumption, and significantly lower than the segregation of face-to-face interactions with neighbors, co-workers, or family members.
The paper calculated an isolation index, which is basically how much people are exposed to a point of view that is similar to their own. An internet where conservatives only read fox news and liberals only read the new york times would have a very high isolation index score. However, because conservative blogs link to the New York Times and liberal blogs link to foxnews.com, that paper finds that the internet has a lower isolation index than national newspapers and the people we interact with everyday.
The problem with this line of thinking is mentioned as a caveat in the paper:
We conclude with an important caveat: none of the evidence here speaks to the way people translate the content they encounter into beliefs. People with different ideologies see similar content, but both Bayesian and non-Bayesian mechanisms may lead people with divergent political views to interpret the same information differently.
This caveat been vastly underrated by most people looking at this research. When reddit.com or the huffingtonpost.com links to a conservative site such as foxnews.com, it is most likely for one of the following reasons:
1. To tell their readers about the lies Fox News is telling their readers
2. To make a liberal point, highlighted with "even Fox News admits..."
3. To make counter arguments to the views expressed, so when the reader visits the website they are inoculated against any of the arguments made.
4. To show their readers that those who disagree are them stupid/racist/misogynous/etc.
As the authors mention, a liberal isn't likely to trust data from a known conservative website in the first place. The same point extends for conservatives linking to liberal websites. A conservative reader who visits an article on nytimes.com because he followed a story about their media bias is still isolated from the liberal point of view.
The paper itself is very interesting, but readers should be wary of drawing the wrong conclusion from this research.