New York Magazine:

Throughout 2017, as social-media-addled minds chased one shiny news object after another, one story was constant and inescapable: “Russia.”

But what “Russia” meant, exactly, depended a lot on where you got your news. The “Russia story” that developed throughout the year was likely very different from the “Russia story” as it was understood by your friends, relatives, or co-workers, not just because your politics might be different than theirs but because you might have been encountering different news entirely. To some Democrats, the “Russia Story” was that Donald Trump was a wholesale asset of the Russian government. To others, the “Russia Story” was the F.B.I.’s methodical investigation. And to many Republicans, the “Russia story” was actually a story about the Democrats’ collusion with Russia, not the Trump campaign’s.

We wanted to see how those competing narratives were shaped, week by week, story by story, through 2017 — examining not just broad “liberal” and “conservative” bubbles, but also the different ways highly partisan readers and their less-partisan neighbors might have encountered the story. Would outlets directed toward zealously partisan Democrats frame the story differently than those with readers more evenly distributed along the political spectrum? What about on the other side of the fence?

To re-create approximate social-media “bubbles,” we used “partisanship scores,” developed by Harvard’s Berkman Klein Center. Rather than attempt to arbitrarily measure the slant of a given publication’s editorial line, these scores gauge the partisanship of its audience, by measuring how frequently its stories were shared by Clinton or Trump supporters during the 2016 election. For our purposes, outlets shared vastly more often by Clinton supporters than Trump supporters are categorized as “Highly Democratic”; outlets where the ratio was closer but still weighted toward Clinton were labeled “Mostly Democratic”; and so on. These reconstructed bubbles aren’t a perfect representation of how people encountered their news, but the partisanship scores allow us to focus on publications’ actual audience — the real citizens of the bubble — rather than our perceptions of their politics.

Fascinating comparative demonstration of the real-world effect of different news sources in a digital age.