My Own Experiment Testing "4 Factors" of Algorithmic Bias

After reading Medium’s article about algorithmic bias, I decided to see how many of the factors I could spot in my own Instagram feed by searching a trending topic about beauty and fashion. 

As background information, I follow very few fashion-centric accounts and influencers. However, the celebrities or public figures I follow do have strong intersections with the world of fashion and beauty. For example, I follow Selena Gomez (actor, also CEO of Rare Beauty), and Koshiba Fuka (actor, also spokesperson of CanMake Beauty). These underlying factors may influence my feed.

I first searched the trending topic, “Coachella outfits,” into my search bar. I selected this topic because it is relevant now, and includes the subject of fashion and beauty. It is also a topic that I have not previously searched before, so it should provide fresh data.

Here is a screenshot of my results:

1. Visual Appeal: It comes as no surprise that all of the posts that I was given are visually appealing. An interesting finding is that one post is art and does not show an influencer. It is important to note that the bias promoting posts of women in bikinis rather than their fully clothed selves does not completely apply here, as Coachella Music Festival happens in the desert, and the fashion culture is usually bikinis and scantily clad. It is also an interesting find that there is a mix of diversity in terms of ethnicity. However, there is a lack of diverse body type and shape. 

2. Behavioral Feedback: These posts are based on my previous engagement with the app and show the algorithm at work. For example, the very first post is that of Vanessa Hudgens, which may be related to my following of Selena Gomez, since both are celebrities of the same generation. The post below that is an influencer that one of my friends follows, which shows that Instagram is using my connections to pick content.

3. Influencer Culture: Two Instagram influencers were prevalent on my feed. One was @bri.constantine, who has 5.6k followers. Compared to other influencers, she does not have much of a following, which makes me curious as to why I was shown her post when I do not follow her. Another influencer is @stxph.h, who has 1.8 million followers. She is followed by three people whom I follow, which could be a big reason why she appeared on my feed. She has significant engagement on her profile, with her Coachella outfit reel reaching 33.9 million views, her highest viewed reel since January 2025.

4. Advertising: Two of the top posts were from businesses, specifically a clothing resale site called @curtsy, and a local boutique called @fashionmilos. Interestingly, neither of these posts has much engagement, with barely any comments each, and the accounts themselves do not have viral traction.

Overall, I would say that this short experiment gave me insight into how certain factors are at play in crafting an engaging algorithm. I would not say that this is definitive evidence that Instagram promotes dangerous beauty standards, but I do see how factors like “advertising” and “visual appeal” are prioritized, and I could see the potential of the "rabbit hole effect" if I continue searching and consuming this topic. This was an interesting experiment to see how my feed is influenced by the factors that contribute to algorithmic bias.