FYI the person who wrote this paper has a long standing hatred of google ever since they warned people his website had malware. His motives might not be entirely academic.
I'm curious about how he measured bias and maybe I missed it in the paper. Anyway he threw out people who communicated via Gmail because they didn't show the same bias, which seems sketchy methodologically. Also his findings could be pretty easily explained without Google doing anything purposefully on their end.
For all the good it did...
/s
"But your honor, I only tried to kill my wife - I didn't succeed!"