Researchers employed by Faceshitte just published this paper in Science, entitled “Exposure to ideologically diverse news and opinion on Facebook”. The basic gist of this paper is the claim that their analysis of ideological slant of what Faceshitte users click on in their newsfeeds supposedly supports that conclusion that it is users’ own ideological inclinations that determines the ideological slant of what they click on more than Faceshitte’s newsfeed algorithms.
I’m not even gonna get into the details of their methodology and the validity of their conclusions from their analyses. This is very well treated by Nathan Jurgenson at Cyborgology, and I recommed his analysis highly. My concern with this paper is more fundamental.
As a scientist, I have a big problem with a legitimate peer-reviewed scientific journal publishing a paper authored by employees of a corporation that describes their analysis of the outcomes of a secret proprietary algorithm controlled by that corporation as if the algorithm is some static, knowable law of nature. When the authors refer to “algorithms”, they are basically lying. Because the day this paper was published, Facebook could have completely changed all of their news feed algorithms, and rendered this paper completely moot.
As scientists, of course we are comfortable dealing with unknowns and even the unknowable as we pursue understanding of complicated natural entities. But these “algorithms” being analyzed by the Faceshitte employees in their manuscript are completely knowable, indeed are known to these researchers’ employer, yet are treated as a black box that can only be studied indirectly by examining its outputs. This is disingenuous in the extreme, and turns the notion of “science” on its head.
I am disgusted at Science for publishing this non-scientific propaganda piece.