The biggest Facebook editorial bias is their user base.
Excellent story from the always-insightful Mathew Ingram on the subject of Facebook and Trending news stories, but in this article (and others I have seen) nobody has mentioned the STRUCTURAL bias Facebook has to be dealing with around Trending topics.
Right now FB has human editors working on Trending, and that introduces one kind of bias. And their algorithms are written by programmers, which introduces another kind of bias. But those algos and human editors are relying on the data produced by Facebook’s USERS…who are NOT demographically representative of the population as a whole, which introduces yet another bias, and one which is likely to be even more important than the other two.
Facebook users are significantly more likely to be women, under 55, urban, connected to broadband and more highly educated. In US terms, all of those demographics skew Democrat rather than Republican (to varying extents.)
In the United States, about 45% of Americans either identify as Democrats or lean that way, compared to 42% leaning or identifying as Republican. Pretty close to a tie. But if I look at Facebook demographics, (especially heavy users who spend the most time and post/share/like the most news stories) I would expect the split to be much wider.
If the data the algorithms and editors are seeing isn’t at least 60% Democrat and 40% Republican I would be surprised. And even a 65/35 split wouldn’t shock me.
[To be clear, this is nothing new. The people in 1980 who wrote letters to the editor for the print version of the New York Times or clipped articles out to share with their friends were almost certainly politically different from the people who did so from the Wall Street Journal. Audience demographics always have skew and bias, and always will. But algorithms make those biases apparent in real time. And no one is talking about that, which seems weird to me.]