"The key may lie in how TikTok's recommendation algorithm - the powerful system that curates the platform's For You Page to predict content you might watch - is managed when it changes hands. Currently, TikTok's system depends on huge amounts of global data and feedback loops, which can change recommendations in an instant. Under the terms of the deal TikTok's algorithm, which will be licensed by investor Oracle, is set to be retrained on American user data."
"Mr Navarra said this could leave the app feeling "safer and sturdier" but also leaving it at a risk of "becoming less culturally essential" as a result. "TikTok's power has always come from feeling slightly out of control - weird, niche, uncomfortable, sometimes politically sharp content for anyone else or before it goes anywhere else," he said. "If you start smoothing those edges, you don't just change moderation. I think you change its relevance.""
ByteDance has signed a deal that places TikTok's US operations under investor-linked control and will license the recommendation algorithm to Oracle. The algorithm is set to be retrained on American user data instead of relying on global datasets. TikTok's recommendation system currently uses vast international data and rapid feedback loops to shape the For You Page. Experts warn that retraining on US-only data may make the app feel safer and sturdier but could reduce cultural relevance and the platform's unpredictable edge. Core features such as short videos and shopping are expected to stay intact, while timely parity of updates will affect user experience.
Read at www.bbc.com
Unable to calculate read time
Collection
[
|
...
]