The Senate Republican Policy Committee’s private meeting with Google did not bode well. after claims of email bias.
The Committee had invited Google’s chief legal officer Kent Walker to discuss a study that found that Google disproportionately filtered Republican lawmakers’ emails into hidden spam folders compared with emails from Democratic lawmakers.
Walker told the committee no bias exists in the way Google handles spam, according to one report.
“The lunch was spirited,” Senator Ted Cruz told Politico. “Google deflected, refused to provide any data, repeatedly refused to answer direct questions.”
Just as search marketers will tell you it takes specific knowledge to target the correct message to gain a click, a share or a lead, email marketers will tell you it’s an art to deliver content into the correct folder.
A recent study of 300,000 emails during the 2020 election that found Gmail was 50% more likely to designate messages from Republicans as spam than those from Democrats sparked the debate.
The paper -- published by researchers at North Carolina State University -- attempts to assess the fairness of spam-filtering algorithms (SFAs) for email services Gmail, Outlook, and Yahoo, as it applied to the 2020 U.S. election from July 1, 2020 through November 30, 2020.
About 102 email accounts were created. The researchers also subscribed to two Presidential, 78 Senate, and 156 House candidate newsletters.
The first experiment studied general trends of biases. The second studied the impact of different email interactions such as reading emails, marking them as spam, or vice versa.
After all, there is a growing body of evidence suggesting biases in online algorithms can influence undecided voters through search queries and results.
One example in the research shows that “bias in search rankings can shift the voting preferences of the undecided voters by as much as 20% without those voters being aware of the manipulation.”
Several U.S. political candidates in the 2020 U.S. election raised concerns that email clients filtered the campaign emails sent to constituents.
In response to the research, Neil Kumaran, Google Group Product Manager of Gmail Security & Trust, published a post. He didn’t respond to the allegations with direct answers, but offered insight into how Gmail works.
He wrote that “some misperceptions circulated regarding how Gmail spam filtering works, but the academics whose research was under discussion have clarified their findings.”
Kumaran went on to explain how Google has built a strong foundation and continues to improve its systems—blocking more than 99.9 percent of spam, phishing, and malware from ever reaching users’ inboxes.
The findings imply that email services mark “emails with similar features from the candidates of one political affiliation as spam while do not mark similar emails from the candidates of the other political affiliation as spam.” The political affiliation of the sender appears to play some role toward the decision of the SFAs.
It’s also possible that SFAs of email services also learnt from the choices of some voters marking certain campaign emails as spam and started marking similar campaign emails as spam for other voters.
“While we have no reason to believe that there were deliberate attempts from these email services to create these biases to influence the voters, the fact remains there that their SFAs have learnt to mark more emails from one political affiliation as spam compared to the other,” researchers wrote.
The email algorithms work similarly to search queries in that they learn from experience.
Researchers found it important for email services to audit SFAs to ensure any “properties of the sender that they consider in determining whether any given email is spam or not are not, unknowingly, putting one side at an advantage compared to the other.”