>"In the weeks leading up to the 2018 election, bias in Google’s search results may have shifted upwards of 78.2 million votes" (that would be 70% of the 113 million people who voted in the election)

You are misinterpreting what he is saying there. That same thing tripped me up too on first glance because the 78.2 million number seemed ridiculously high. He is referring to votes and not people who voted. If there were 10 items on your ballet and he thinks Google may have influence you, he is counting that as 10 votes and not 1 person voting. There is nothing technically wrong with displaying the number that way, but it seems so misleading that it makes me question if it was intentionally chosen to make the problem sound worse. It is decisions like that which make me doubt his neutrality in presenting the data.