TikTok is an endless source of content. When a person starts watching a video, they just have to slide their finger across the screen to get to the next one. The dynamic is repeated over and over again until the person realizes that they have spent hours watching videos on the platform.
This platform is not defined as a social network, but as an entertainment place where users can find content that is valuable to them, but how does it know users so well to deliver videos that are meaningful to them?
The four goals of the TikTok algorithm
According to an internal TikTok document accessed last December, there are four priorities for the platform’s algorithm: user value, long-term user value, creator value, and platform value.
According to the company’s spokeswoman, Hilary McQuaide, that document was authentic and worked to explain how the platform’s algorithm works, which has become the main vehicle of culture for young people.
To fill users with such valuable content, the platform takes into account various factors, including user likes and comments, as well as the information of videos they consume, such as subtitles, hashtags and sounds. that are used in them.
The document also revealed that “retention” and “time spent” are two of the most important metrics it takes into account, because the application wants people to spend most of their time on it through content that they like. seems satisfactory.
How does TikTok decipher its users?
However, the viewing time of a video is not the only factor to consider, because to understand users, the platform also makes use of machine learning with which it performs an equation between likes, comments, time of playback and an indication that the video was viewed.
“The recommendation system scores all videos based on this equation and returns the videos with the highest scores to users,” it was explained in the document to which NYT had access.
However, based on this learning, the platform also decides not to deliver similar content all the time, since if it did, the user would end up getting bored and closing the application. Therefore, they decide to disperse the content that is delivered to them.
Although for some specialists these descriptions are reasonable, but traditional in data processing through artificial intelligence, the most outstanding point is that the platform has an enormous volume of data to feed on and continue learning, in addition to the fact that users are “highly engaged”
The goal of TikTok, instead of creating an addiction among users, is to add more people daily. Julian McAuley, a computer science professor at the University of California, told the US media that there is no magic code behind the platform by which people get hooked.
Although these systems are quite complex, the outlet noted, “there are no inherently sinister or incomprehensible elements about the TikTok recommendation algorithm described in the document.”