The content recommendation algorithm, which powers the short video platform TikTok, has come back into the spotlight after the Chinese company ByteDance, the app’s owner, signed binding agreements to establish a joint venture that grants it management of TikTok’s application operations in the United States to American and global investors, including the cloud computing company Oracle.
Here’s what we know so far about the fate of this algorithm after the joint venture was established.
Will ByteDance relinquish control over the algorithm?
While establishing this new entity is a significant step towards avoiding a ban on the TikTok application in the United States, as well as alleviating trade and technological tensions between Washington and Beijing, there is still ambiguity surrounding the ownership of the recommendation algorithm, which is considered the crown jewel of TikTok.
Rush Doshi, who served in the National Security Council under US President Joe Biden, stated on platform X that it’s unclear whether the algorithm has been transferred, licensed, or is still owned and controlled by Beijing, with Oracle’s role limited to "monitoring" only, according to Reuters.
The algorithm is a key element in TikTok’s global success, and even a few months ago, ByteDance’s position was that it would prefer to close the app in the United States rather than sell it.
But in September, Reuters reported, citing sources, that ByteDance would retain ownership of TikTok's commercial operations in the United States but would cede control over the app's data and content and its algorithm to the joint venture.
According to the sources at the time, the joint venture would act as a backend operating platform for the American company, managing user data in the United States and the algorithm, while a separate unit, wholly owned by ByteDance, would continue to control revenue-generating business activities such as e-commerce and advertising.
Two informed sources told Reuters on Friday that these arrangements formed the general framework of the deal announced on Thursday, and they clarified that the "TikTok United States" entity under ByteDance's control would remain the revenue-generating body, while the new joint venture would receive a portion of these revenues in exchange for its technical and data services.
The Chinese government has not yet announced its stance on the signed deal. Beijing had made amendments to export laws in 2020 that grant it rights to approve any export of algorithms or source codes, adding a layer of complexity to any attempt to sell the app's business in the United States or separate it from the parent company.
What is the secret behind the algorithm’s strength?
Analysts say that TikTok’s global success lies not only in its algorithms but also in how they work alongside the short video format.
But TikTok has proven that the algorithm based on understanding user interests can be more effective. Instead of building its algorithm on the "social graph" like Meta, TikTok executives have clarified that the platform's algorithm relies on "interest indicators".
The short video format allows TikTok’s algorithm to be more dynamic and capable of tracking changes in users' preferences and interests over time, achieving extremely precise levels of detail, such as what a user might prefer at a specific time of the day.
In addition, TikTok was designed as a mobile-first application from the start, giving it a competitive edge over competing platforms that had to adapt their interfaces from computer screens.
Moreover, TikTok's early entry into the short video market gave the company a first-mover advantage; while Meta launched the short video feature "Reels" on Instagram only in 2020, Alphabet launched the "Shorts" feature on the YouTube platform in 2021, both lagging behind TikTok in terms of years of accumulated data and product development experience.
What do studies reveal about the algorithm?
TikTok regularly recommends content that does not fall within the known interests of users, a fact that the company’s management has repeatedly emphasized as important for the user experience.
A study published last year by researchers from the United States and Germany concluded that TikTok’s algorithm "exploits user interests in 30% to 50% of the suggested videos", after analyzing data from 347 users and five automated programs.
The researchers wrote in a paper titled "TikTok and the Art of Customization": "This result suggests that TikTok’s algorithm prefers to suggest a large number of exploratory videos, either in an attempt to better understand user interests or to increase its retention rate by suggesting many videos outside of their (known) interests".
