This site is fictional demo content. It is not real news or affiliated with any real organization. Do not treat it as fact or professional advice.

NextPaper Logo
nextpaper.ccCYBER CHRONICLE
RSS
EN中文

Full article

FULL TEXT

View this issue
Deep diveINTERNET

One Year of Algorithm Governance: What Platforms Did, What Users Got

In April 2027, China's网信办 issued algorithm recommendation regulations requiring platforms to offer 'turn off personalization' options. One year later, this report evaluates the governance effects: progress is clear, but structural dilemmas remain.

Content

One year after algorithm recommendation management regulations, how effective have they been?

Changes on the User Side

According to a third-party evaluation report published in March 2028, mainstream platforms (Douyin, Kuaishou, WeChat, Bilibili, Xiaohongshu) have all implemented "turn off personalization" options, allowing users to switch to non-personalized recommendations with one click. After turning off, recommendation content diversity index improved by approximately 23% on average, with some alleviation of information cocoon effects.

However, actual usage rates are remarkably low: surveys show less than 40% of users know this function exists, and among those who know, only 12% actually turn it off. Users generally反馈 "recommendations got worse after turning off," indicating current non-personalized recommendation algorithms still have significant room for improvement.

Changes on the Platform Side

Platform changes are equally significant. Major platforms have all launched "algorithm transparency reports," disclosing basic frameworks of recommendation logic; timeline and social recommendation weights have increased while pure algorithm-driven immersive recommendations have decreased; teen modes have substantially improved in functionality, no longer being "simplified versions."

But critics point out platform changes are more "defensive compliance" than proactive transformation. The commercial value of personalized recommendations is too high—with equal traffic, personalized recommendation ad conversion rates are 2-3x non-personalized—the platform lacks endogenous motivation to truly weaken personalized recommendations.

Unresolved Structural Dilemmas

After one year of algorithm governance, the industry has gradually recognized several fundamental challenges:

Information cocoons cannot be solved by "turn off button" alone. Even after turning off personalized recommendations, the overall content ecosystem is still driven by traffic mechanisms—eye-catching content naturally gets more exposure than in-depth content; algorithms only accelerated this process rather than created it.

User autonomy awareness and actual experience have gaps. Many users stated "I know the algorithm is influencing me, but I still find good stuff through algorithm recommendations"—this contradictory mentality reflects the real dilemma of users in the information overload era: completely autonomous information selection itself is high-cost cognitive labor.

Cross-platform data connectivity has no consensus yet. Users turn off personalization on Platform A but are still tracked on Platform B, with the two systems often backed by the same company's data center. True privacy protection requires coordination mechanisms across platforms and even regulatory levels.

Boundary

This is fictional content for entertainment only.