âThis system means viewing time is key. The algorithm tries to make people addicted rather than giving them what they really want, âsaid Guillaume Chaslot, founder of Algo Transparency, a Paris-based group that has researched YouTube’s recommendation system and has a gloomy view of the effect of the product on children, in particular. Mr. Chaslot reviewed the TikTok document at my request.
âI think it’s a crazy idea to let TikTok’s algorithm rule our children’s lives,â he said. âEvery video that a child watches, TikTok gets information about him. In a few hours, the algorithm can detect his musical tastes, his physical attraction, if he is depressed, if he is on drugs, and many other sensitive information. There is a high risk that some of this information will be used against him. This could potentially be used to micro-target him or make him more addicted to the platform. “
The document states that viewing time is not the only factor that TikTok considers. The paper offers a rough equation for how videos are rated, in which a prediction based on machine learning and actual user behavior is summarized for each of the three data bits: likes, comments, and rating. playing time, as well as an indication that the video has been played:
Plike X Vlike + Pcomment X Vcomment + Eplaytime X Vplaytime + Pplay X Vplay
âThe recommendation system assigns scores to all videos based on this equation and returns the videos with the highest scores to users,â the document said. âFor the sake of brevity, the equation presented in this document is very simplified. The actual equation used is much more complicated, but the logic behind is the same.
The paper illustrates in detail how the company is fine-tuning its system to identify and remove ‘like bait’ – videos designed to play with the algorithm by explicitly asking people to like them – and how the company is thinking about questions. more nuanced.
âSome authors may have cultural references in their videos and users can only better understand these references by watching more of the author’s videos. Therefore, the total value that a user watches all of those videos is greater than the values ââof each video added up, âthe document states. âAnother example: if a user likes a certain type of video, but the app keeps pushing the same genre, they would quickly get bored and close the app. In this case, the total value created by the user watching the same type of videos is less than watching each video because the repetitiveness leads to boredom.
âThere are two solutions to this problem,â the document continues. âMake assumptions and break down value into a value equation. For example, in terms of repeated exposure, we could add a value ‘same_author_seen’ and for the boredom problem, we could also add a negative value ‘same_tag_today’. Other solutions besides the value equation may also work, such as forced user recommendation for feed and scatter etc. For example, the problem of boredom can be solved by dispersal.