Key Points
- China’s Cyberspace Administration (中央网信办) is directing major platforms like Douyin (抖音) and Xiaohongshu (小红书) to rework algorithms due to issues like “information cocoons” and polarizing content.
- Platforms have signed the Nanning Declaration (南宁宣言) and are increasing algorithm transparency by publicizing rules and principles.
- New features like “Cocoon Assessment” and “One-Click Break the Cocoon” are being implemented to help users diversify their recommended content.
- Platforms are enhancing content review mechanisms to promote positive content and prevent the recommendation of vulgar or undesirable information.
- Governance of algorithms is a long-term project, with regulators conducting inspections and supervising platforms to ensure continuous improvement in content quality and user empowerment.
The challenge of algorithm recommendation chaos is prompting significant action across China’s digital landscape, with major platforms now under the microscope to refine how content reaches users.
It all kicked off with the “Clear and Bright · Algorithmic Issues Governance of Online Platforms” special action, a directive from the Cyberspace Administration of China (Zhongyang Wangxin Ban 中央网信办).
This initiative is pushing key online platforms to seriously rethink and tweak their information recommendation algorithms.
Why the big push? Netizens have been vocal about some pretty pressing issues.
We’re talking about algorithms that sometimes promote vulgar information, create stronger “information cocoon” effects (where you only see what you already agree with), and even ramp up viewpoint polarization.
Sounds familiar, right? These are global challenges for any platform using recommendation engines.
So, how are China’s tech giants responding?
They’ve actively stepped up, notably by signing the Nanning Declaration (Nanning Xuanyan 南宁宣言) on “Algorithms for Good.”
This isn’t just talk; they’re rolling out changes.
Platforms are beefing up content review for what algorithms suggest.
They’ve launched dedicated features or even accounts to publicly spill the beans on their algorithm rules and principles – a big step for transparency.
Innovations like “Cocoon Assessment” and “One-Click Break the Cocoon” features are emerging, designed to give users more control and diversity in their feeds.
Plus, they’re enhancing services for managing user interests and preferences, aiming to diversify the content that gets recommended.
Recently, a lineup of major players, including Douyin (Douyin 抖音), Xiaohongshu (Xiaohongshu 小红书), Weibo (Weibo 微博), Kuaishou (Kuaishou 快手), WeChat Video Account (Weixin Shipinhao 微信视频号), and Bilibili (Bilibili 哔哩哔哩), have systematically optimized and improved multiple functions.
Their focus? Key areas like:
- Weighted recommendation of positive content.
- Ensuring user autonomy.
- Optimizing recommendation content diversity.
- Enhancing algorithm transparency.
Shedding Light on Algorithms: Boosting Transparency Initiatives
A core part of this overhaul is publicly disclosing how these complex algorithms actually operate, ensuring users have a right to know what’s shaping their online experience.
Take Douyin (Douyin 抖音), for example.
They’re explaining their recommendation logic, intervention mechanisms, and governance outcomes through their “Safety and Trust Center” website and even hosting open day events.
This level of openness can build significant user trust.
- Douyin (抖音): Explains logic, intervention, and outcomes via “Safety and Trust Center” and open days.
- Weibo (微博): Increased Hot Search transparency by revealing ranking and data rules, launched popularity tags.
- WeChat Video Account (微信视频号): Uses accessible guides (text/image/video) like “Understand WeChat Video Account Algorithm Recommendations at a Glance” and “Algorithm Breaking the Cocoon Series”.
Weibo (Weibo 微博) is focusing on its influential Hot Search feature.
They’ve increased transparency by publicly revealing ranking rules and data rules.
They’ve also launched Hot Search popularity tags to show the driving factors behind trending topics. Understanding why something is trending is crucial for media literacy.
WeChat Video Account (Weixin Shipinhao 微信视频号) is using easy-to-understand text-image and video content to inform users.
They’re publishing guides like “Understand WeChat Video Account Algorithm Recommendations at a Glance” and an “Algorithm Breaking the Cocoon Series.”
Making complex systems digestible is key for broad user understanding.
Resume Captain
Your AI Career Toolkit:
- AI Resume Optimization
- Custom Cover Letters
- LinkedIn Profile Boost
- Interview Question Prep
- Salary Negotiation Agent
Escaping the Echo Chamber: New Tools to Break “Information Cocoons”
One of the biggest criticisms of modern algorithms is their tendency to create “information cocoons,” where users are only exposed to confirming viewpoints.
Platforms are now innovating with features like “Cocoon Assessment” and “One-Click Break the Cocoon” to help users mitigate this risk.
Douyin (Douyin 抖音) has comprehensively upgraded its “Usage Management Assistant.”
It’s introduced an innovative content preference assessment function that visualizes recent browsing content, giving users a clear picture of their digital diet.
Visual feedback can be a powerful tool for self-awareness.
Xiaohongshu (Xiaohongshu 小红书) has set up “Content Preference Assessment and Adjustment” and “Explore More” functions.
These allow users to easily browse more diverse recommended content with a single click.
The “one-click” solution lowers the barrier for users to explore beyond their usual bubble.
Kuaishou (Kuaishou 快手) is leveraging its positive energy algorithm.
The goal is to increasingly discover and fully present content that is “positive,” “useful,” “warming,” and “trustworthy” in its algorithm recommendations.
This deliberate push towards positive content is a significant directional shift.
Raising the Bar: Improving Recommendation Content Review Mechanisms
It’s not just about transparency and breaking cocoons; it’s also about the quality of what’s being recommended.
Platforms are continuously improving their recommendation content review mechanisms.
There’s a strong emphasis on strengthening the push of positive energy content and preventing algorithms from recommending vulgar or undesirable information.
WeChat Video Account (Weixin Shipinhao 微信视频号) has improved its dual mechanism of “friend recommendations” and “algorithm recommendations.”
They’re constantly iterating and upgrading their identification and crackdown models.
They’re also strictly prohibiting typical undesirable information like vulgar or distasteful content from even entering the recommendation pool.
A dual mechanism can offer a more balanced and human-centric recommendation approach.
Douyin (Douyin 抖音) has introduced an innovative mechanism for verifying hot topic participants.
This is aimed at preventing malicious propagation behaviors such as staged acts, mimicking trends for clicks, and piecing together edited content for misinformation.
Tackling the authenticity of trending content is a critical, and often overlooked, aspect of platform governance.
Find Top Talent on China's Leading Networks
- Post Across China's Job Sites from $299 / role, or
- Hire Our Recruiting Pros from $799 / role
- Qualified Candidate Bundles
- Lower Hiring Costs by 80%+
- Expert Team Since 2014
Your First Job Post
Putting Users in Control: Enhancing Empowerment Features
Ultimately, many of these changes are about giving users more direct control over their feeds.
Platforms are continuously optimizing features for interest preference management and negative content feedback.
The aim is to make it easier for users to independently adjust and refine the content algorithms recommend to them.
- Kuaishou (快手): Provides detailed interest preference management with adjustable intensity via sliding tags.
- Weibo (微博): Offers specific negative feedback options (“not interested,” “don’t see this blogger,” “low quality”) for accurate response.
Kuaishou (Kuaishou 快手) provides users with convenient and detailed interest preference management functions.
Users can adjust the push intensity of different content based on their preferences by simply sliding corresponding interest tags.
Granular control like this can significantly improve user satisfaction.
Weibo (Weibo 微博) offers various negative feedback options for users to choose from.
Options include “not interested,” “don’t see this blogger,” and “low content quality,” allowing the platform to accurately respond to user needs.
Specific feedback options are far more useful than a generic “dislike” button.
A relevant official from the Cyberspace Administration of China (Zhongyang Wangxin Ban 中央网信办) provided a candid perspective.
While acknowledging that certain results have been achieved in the governance of information recommendation algorithms on these key platforms, they also noted that issues still exist.
The effectiveness of some functions might be limited, and the quality of recommended content might still be insufficient.
There’s still a gap, they stated, compared to the expectations of netizens and the wider society.
This isn’t a one-and-done fix.
Algorithm governance is seen as a long-term, systematic project.
Cyberspace administration departments will conduct regular inspections.
They will supervise information recommendation algorithm platforms to continuously optimize algorithm operating mechanisms and management rules.
The ongoing goals are to constantly improve the quality of recommended content, explore diverse paths for “breaking cocoons,” innovate user empowerment functions, and effectively safeguard the legitimate rights and interests of netizens in the evolving landscape of algorithm recommendation systems.
ExpatInvest China
Grow Your RMB in China:
- Invest Your RMB Locally
- Buy & Sell Online in CN¥
- No Lock-In Periods
- English Service & Data
- Start with Only ¥1,000
![China's CAC Launches Qinglang Campaign- Cracking Down on AI Technology Abuse [FreshFromChina]](https://freshfromchina.com/wp-content/uploads/2025/05/Chinas-CAC-Launches-22Qinglang22-Campaign-Cracking-Down-on-AI-Technology-Abuse-FreshFromChina-150x150.png)
