TikTok’s “Pretty” Algorithm (Take 2)

Rani Woerner
SI 410: Ethics and Information Technology
3 min readFeb 27, 2021

I don’t know about you, but I’ve spent countless hours scrolling through the infinite stream of videos on TikTok’s “For You” page, a page filled with engaging videos placed there by TikTok’s algorithm based on what content it thinks I would enjoy. But while TikTok’s algorithm has successfully entertained me for hours on end, it is also riddled with biases, allowing only a select few to achieve success on the platform.

According to TikTok, every video posted has an equal chance of “going viral” and popping up on people’s “For You” pages, but based on my own experiences using the app, this claim evokes suspicion. Almost all of the videos on my “For You” page feature stereotypically attractive people. One day in particular, as I was scrolling through my “For You” page, I saw video after video of physically flawless women, and as I started to feel badly about my own appearance, I knew something was wrong. While it would make sense that many of the app’s most popular creators would be physically attractive (think celebrities), it seemed odd that my “For You” page looks like it was taken over by models if it were in fact true that all videos have an equal chance at success.

In her book Algorithms of Oppression, Safiya Noble asserts that there are oftentimes biases present in algorithms and artificial intelligence, and in the case of TikTok, she is absolutely right. Journalist Katie Mather reports that recently leaked documents from TikTok indicate that a creator’s exposure might be limited if they are deemed “unattractive.” To determine attractiveness, TikTok uses an algorithm that takes certain measurements of the creator, evaluates the degree of fit with ideal averages, and provides what is known as a “beauty score.” Mather provides a list of specific attributes that are assessed and could cause a creator’s video to get flagged and filtered out from the “For You” page. This bias against creators who don’t fit modern beauty standards prevents masses of people from viewing their content, making it nearly impossible for their videos to succeed.

An example of the criteria algorithms use to measure attractiveness.

Noble also discusses the implications of biases in algorithms and how harmful they can be, writing that “these human and machine errors are not without consequence.” In this case, Noble is spot on; TikTok’s beauty algorithm is by no means consequence-free. Aside from creators who are deemed “unattractive” not being able to share their content, messages, and stories, having a “For You” populated mostly by extremely attractive people can have detrimental effects on the mental health and body image of the younger audience that the app attracts.

In short, TikTok’s “pretty” algorithm needs to go.

--

--