New Study How Algorithms Shape the Distribution of Political Advertising on Social Media Platforms
Everything You Need To Know In Less Than 50 Words
A new study published by the Arxiv lab at Cornell University has illuminated that social media platforms have failed to represent how they distribute political campaign ads. Social media platforms play a large part in political advertising, with targeted ads on the platforms being a major spend for most campaigns.
Tell Me More
Online platforms such as Facebook, Google, and Tik Tok play a massive role in disseminating political news and information to the general public. They also play a large part in distributing political ads.
During the 2020 presidential election, social media giants created ad libraries to collect and track political ads run by the platforms. Platforms did this to bring greater transparency to how their political advertising algorithms distribute political ads across the platforms. It was also part of an effort by social media platforms to stave off regulations such as the Honest Ads Act.
The Study’s Procedure
The study analyzed over 800,000 data points from Google and Facebook ad libraries from the two weeks prior to the 2020 presidential election. They also examined 2.5 million Tik Tok videos from the same period.
Researchers evaluated the data based on three criteria:
- Did the data disclosed meet the platform’s self-imposed objective of holding political advertisers accountable?
- Did the data disclosed by the platforms meet the same criteria and standards that radio and TV broadcasters are held to?
- Did the platform disclose everything they know about how their advertising algorithms distribute the content, who the ads are for, and the ad targeting criteria?
The Study’s Findings
The study found that Google, Facebook, and Tik Tok failed to meet the three criteria requirements. In the researchers’ own words, “the efforts of the major platforms to make the necessary disclosures to understand their practices fall woefully short.” According to the findings, the ad libraries maintained by these platforms do not meet any of the objectives.
The study found that the data only partially disclosed the ad targeting criteria and intended audiences for political ads. The disclosures did not allow the researchers to understand how political campaigns had reached prospective voters and how algorithms affect media consumption.
Secondly, researchers found that social media platforms offered advertising space at different rates to different campaigns. The Trump campaign paid more per impression than the Biden campaign for Facebook political advertising. On Google political ads, it was the other way around. Meanwhile, TV and radio broadcasters are required to offer political advertising space at a fixed rate. They must also offer political ad space at the same price as regular commercial ad space.
Thirdly, researchers found that the platforms had utilized new and untracked political advertising methods, such as using influencers to promote political content. This type of advertising remains unregulated. It is impossible to know how impressions generated by influencers affect political campaigns.
It appears that the ad libraries maintained by social media platforms fall woefully short of achieving the goals that the platforms themselves set out to achieve. Researchers indicate that we need more comprehensive disclosures from platforms to understand the role of political advertising algorithms in political campaigns.