China's Algorithm Oversight: A Model for the US and EU?
Written on
Understanding China's Algorithm Scrutiny
China's data regulatory body, the Cyberspace Administration of China (CAC), has unveiled a list of 30 algorithms from leading tech companies, including Alibaba and ByteDance, for examination. This follows the introduction of new regulations designed to enforce stricter oversight on these tech giants.
CAC mandates that these firms conduct a self-assessment regarding the security of their algorithms, the data collected—which may include sensitive biometric or identity information—and the sources of data used for algorithm training. This recent development opens the door for CAC to potentially evaluate the internal processes of these corporations.
The public list of algorithms released by CAC offers only a superficial overview of the companies' functions. For example, it notes that Weibo's recommendation system surfaces content based on users' previous browsing activities. While this information may seem obvious, the deeper implications are more significant.
What remains unclear is the extent to which Douyin (the Chinese counterpart of TikTok, also owned by ByteDance) leverages user data to power its recommendation system. The specifics of how it matches users with particular videos and the complete range of data sources utilized are still a mystery.
This initiative may not clarify things for external observers, but it will empower the government to scrutinize how these platforms make automated decisions. They are likely to have numerous inquiries and could even involve engineers in discussions about their algorithmic design processes. Future regulatory actions will likely reflect CAC's findings.
It’s important to note that these tech companies would not willingly submit such detailed information. They are likely to safeguard their closely held business secrets, similar to how Meta has managed to protect its algorithms in the US under scrutiny.
The broader narrative here revolves around the dynamics between business and government in the world's largest economies. Both China and the US are examining social media platforms, though with differing motivations.
Why is China Taking Action?
Officially, the rationale behind these measures is attributed to "data abuses" by tech companies. CAC has previously levied significant fines for data privacy violations, such as the $1.2 billion penalty imposed on ride-hailing service DiDi earlier this year. However, this recent action seems to aim at a more comprehensive understanding of the relationship between technology services and consumers.
On another level, it’s plausible that China is responding to the detrimental impacts of social media observed in Western nations. The Chinese government likely realizes that these platforms can influence political opinion, leaving citizens vulnerable to external manipulation—something a one-party state would want to control.
Moreover, should the Chinese government gain insight into how these algorithms function, the potential for enhancing their own propaganda efforts is substantial. This knowledge could serve two primary purposes:
- Maintaining Control Domestically: China could amplify its own messaging while suppressing dissenting views, thereby tightening its grip on the tech sector.
- Exploiting Social Media Abroad: Understanding the operational logic behind major tech algorithms could enable China to influence trends and narratives outside its borders.
Reports have indicated that China has already begun implementing such strategies. For instance, Buzzfeed revealed that former ByteDance employees alleged that the company directed staff to embed pro-China content within the US news app TopBuzz.
The simplicity of the TopBuzz initiative is striking.
Of course, there remains the possibility that China could leverage social media to exacerbate divisions within Western countries, although that currency is waning.
This scenario may sound alarming, and perhaps it should, but it reflects the historical patterns of foreign policy. In his book, Empire of Ideas: The Origins of Public Diplomacy and the Transformation of U. S. Foreign Policy, Justin Hart discusses how the US State Department used cultural figures, like Louis Armstrong, as tools of propaganda in the 1960s.
In today's context, TikTok could serve a similar role.
ByteDance claims that TikTok operates independently with its headquarters in Singapore and minimal interference from Beijing. The platform has shifted its operations to Oracle's cloud servers and has requested an audit of its recommendation algorithms by Oracle.
The US, however, remains skeptical of these assertions, with many lawmakers worried that ByteDance employees may still access US citizens' data. Additionally, EU nations have ruled that their citizens' data cannot be processed in the US, making it unlikely that EU data will find its way to ByteDance’s servers when the legal challenges arise.
China has learned valuable lessons from the US's cultural hegemony in the 20th century. It is likely to maintain the façade of separation between state and business for the sake of its long-term objectives.
The Evolution of Social Media Dynamics
Facebook experienced explosive growth thanks to its social graph; now, TikTok's content graph represents the next evolution in social media.
China's desire to control censorship is increasingly challenged by the complexities of opaque algorithms, live-streamed content, and instantaneous interactions. Although we may have reached a critical juncture, the state is eager to act before it becomes too late. China possesses the capability to wield this power, even as the US has chosen a different path.
As previously mentioned, social networks are transitioning away from connecting users to their friends’ content and instead are adopting TikTok-style algorithms that deliver content based on user engagement predictions. The objective is simple: to keep users engaged at all costs. Platforms ranging from Facebook to Twitter and even Amazon are now attempting to replicate TikTok's model.
Earlier this year, a leak revealed that TikTok shares an internal document titled "TikTok Algo 101" with select departments. Although it does not disclose any significant trade secrets, it outlines the company’s objectives: "The company’s 'ultimate goal' is to increase daily active users by boosting user retention rates and total time spent on the app." Each video receives a score based on likes, comments, and viewing duration, emphasizing a formula for success.
In this framework, mere celebrity status no longer guarantees social media engagement. Anyone can achieve virality by producing content that partially meets audience preferences, making the platform enticing for both content creators and consumers.
In a landscape where attention is up for grabs, and "quality" is subjective, AI programs are increasingly seen as ideal content creators. Many brands in China are turning to virtual influencers for their live shopping broadcasts, as they can create avatars tailored to different audience segments. With appropriate data inputs, this technology could generate countless variations on content themes, which algorithms would swiftly deliver to the right users.
A report by Media Matters for America highlighted how TikTok can lead users down a "rabbit hole" of increasingly extreme content, often beginning with benign topics like fitness and gradually introducing more radical viewpoints. There is little evidence to suggest that the algorithms were designed with such outcomes in mind, underscoring the need for regulation.
Protecting Users: Divergent Perspectives
Opinions vary on the best methods for safeguarding users. One perspective argues that as long as the outcomes are consistent with the inputs, the system is functioning correctly. In other words, if users consent to share their data and are satisfied with the results, everything is fine—akin to ordering a hot dog and enjoying the taste, regardless of the preparation method.
Tech companies will likely advocate for this viewpoint to protect their intellectual property for as long as possible. They may agree to a degree of oversight, potentially from an external entity of their choosing. However, it’s evident that this approach will be inadequate since these firms thrive on maximizing user engagement to support their advertising models.
Thus, when considering whether the US and EU should adopt China's approach, we must clarify our desired outcomes and reverse-engineer from there. It remains uncertain whether increased transparency from tech giants will provide the state with the insights necessary to implement effective reforms.
I recall a compelling Twitter thread from a former Reddit CEO discussing the challenges of eradicating misinformation from the platform. Each well-meaning policy tends to give rise to new problems.
Addressing such a pervasive challenge requires a comprehensive solution that encompasses psychology, behavioral science, socioeconomics, politics, statistics, platform design, artificial intelligence, and data management.
While significant advancements have been made in each of these areas, the question remains: who will integrate them into a cohesive solution? Tweaking the algorithm is akin to using a scalpel when a sledgehammer is needed.
Ultimately, the motivations of both China and the US in this context align closely. Both nations seek to exert greater control over big tech companies and manage the content accessible to their citizens. However, their methods will differ significantly due to the distinct underlying intentions of each state.
The US could take a page from China's playbook by proactively pursuing its objectives. Given the potentially harmful nature of some of China’s goals, this urgency should drive the US to take action.
Algorithms play a crucial role in the dynamics—both positive and negative—of social networks. Yet, they are not inherently "evil"; any malicious intent originates externally. Effective regulation must consider the entire ecosystem. Scrutinizing algorithms is merely the first step.