[charlesarthur]

The NBER paper that you point to in a subsequent comment doesn't have any detail on the popularity of social media; we explicitly can't make any determination on whether social media is having an effect there because the data is missing. If, say, Facebook were to provide external researchers with data about the growth in Facebook use - users, median time spent on site, average time spent, SD of time spent - for a number of years, that would help to identify whether social media has a role in such polarisation.

Although it's trivial to say "I see a lot of polarisation on social media, therefore it's worse than it was", a satellite-view paper like that NBER one gives zero insight into the role of social media, partly because the data isn't provided, but also because it doesn't examine what effects there might be on smaller groups within the population who are, say, heavy social media users.

I think the most useful thing Facebook could do would be to make more information available to researchers, rather than pointing to research which hasn't been able to use data and claim that helps exonerate Facebook.