Political Biases in Email Spam Filtering Algorithms During US Election 2020

Someone posted a more casual article describing this not too long ago; I'll summarize the same response I had to that here. Spam filters learn what counts as spam based on what its users, en masse, mark as spam. If a filter is blocking a specific party's emails more, it's because the user base marks that party's emails as spam more. In other words, the spam filter is doing its job: its users have said they, in general, don't want that party's emails, so it blocks them. The difference in seeming bias across platforms is because there's a difference in demographics in the user bases of different services (markedly, for instance, by age).

In other words: spam filters block based on what their users don't want to see, so if you don't want your political emails blocked, stop sending shit the users don't want to see.