You may have already experienced this. When you spend time on TikTok, you quickly realize that the algorithm knows you a little too well. It can even be confusing because its recommendations don’t match old Google searches or interactions we may have had on the platform.

So, how does the Chinese social network manage to stick to our tastes? That’s the question the Wall Street Journal asked itself when it conducted its own experiment to find out more. Even before starting, the American media has its own idea. It seems indeed that the algorithm observes the time spent by a user on a specific content before it moves on to another.

In just a few minutes, TikTok identifies your tastes

In concrete terms, the journalists deployed over 100 fake automatic accounts on TikTok. These bots then watched hundreds of thousands of videos every day. Our colleagues cite the example of this fake profile of a 24-year-old man from Kentucky: kentucky_96.

Very little information is given about him, apart from his IP address and a date of birth. No concrete data on his gender or his interests is present, but this does not prevent the algorithm from hitting the bull’s eye in just a few minutes. For example, kentucky_96 favored sad and depression-related content, and so he focused on a video about a relationship with another person with the hashtags sad and broken heart.

Very quickly, TikTok offers him videos on breakups except that the bot is not really interested, and then lingers on the themes of mental health, depression and anxiety. The algorithm corrects its mistake almost immediately and kentucky_96 receives 93% of videos related to these topics.

See also  LG TONE Free renews its wireless headphones TWS

This figure is very high compared to YouTube, which offers about 70% of content via its recommendation system. For the remaining 7%, a TikTok spokesperson explains that they allow the user to leave his comfort zone. However, our colleagues were able to see that it was mainly advertising.

In the past, TikTok had already returned in detail on the functioning of its algorithm, you can also find our article on this subject here.

Tik TokTik Tok