I scrolled on TikTok to laugh and ended up crying. The algorithm knows you a little too well…

A few nights ago, my bedtime ritual was beginning, and expecting to have some smiles and laughs before sleep time, I ended up crying after watching an extremely accurate 8-second video that caught me off guard. This was followed by an almost pitty-laugh at myself, when I realized that my TikTok (the fact that I am calling it “my TikTok” deserves another one of those pity-laughs) knows me so well that it even seems to know those weak spots in my heart. 

(Here is the video, just for extra-extra-extra context)

I have always been a defender of the app because honestly, it never failed to bring me some sort of really accurate comfort that some days was much needed. While many of my friends were deleting the app to spend less time online, I disagreed stating that I did think that it was possible to create healthy habits while still obtaining those “good” feelings, in the end, it was just a dopamine rush that lasts as long as the video. 

Anyway, after my little crying incident, I felt the opposite of comfort but most importantly it made me want to know why the algorithm knows me too well to even throw some very accurate crying material here and there. This blog is dedicated to that, and a little introduction (to myself and others) to quite basic -but scary- algorithms. 

TikTok’s algorithm works under a simple structure named user-item matrix. It is pretty much a table that in one column has the “individual users” (me) and in the other one, an “individual item” that the app needs to recommend to the user (the video that made me cry). Then, there is the value aspect in which the so-called “feedback” is meant: how did the user interact with the item? And so the chain begins. There is the Implicit feedback that translates into commonly unintentional behavior such as clicking for a few seconds and the Explicit feedback which is based on more specific interactions such as “rating” or commenting. Finally, there is the Granular feedback which simply relates to how much information the combined feedback gives to the platform on a scale from 1-10. Here comes the trick; with TikTok, the sole act of scrolling and the amount of time spent on each video is a form of granular implicit feedback that is also continuous. Simultaneously, because the format of the app is so “easy to digest”, the user is very likely to watch more. In that manner, the feedback becomes bigger and the recommendations, more accurate. 

And apparently, I am not the only one who feels so intrigued by the closeness of TikTok’s algorithm and myself because every other article headliner that I stumbled upon when researching, started with the statement  “The algorithm knows me better than I know myself.” So the more we consume, the more we will get to know ourselves and the more surprised we will continue to get.

I still think it is secretly listening to my thoughts…