While improving user experience and increasing platform stickiness, the algorithms of social media platforms may exacerbate social divisions through a variety of mechanisms. The following are the main reasons and mechanisms: 1. **Filter Bubbles**: - **Definition**: Algorithms push personalized content based on users' interests, browsing history and interactive behaviors, so that the information users are exposed to is highly consistent with their existing views. - **Impact**: Users are in a similar information environment for a long time and lack diversified information sources, which leads to deepening cognitive biases and difficulty in understanding and accepting different opinions. 2. **Echo Chambers**: - **Definition**: The information within the community where users are located is highly consistent, which strengthens existing beliefs and prejudices and reduces exposure to different opinions. - **Impact**: Opinions within the group converge, the views of external groups are ignored or belittled, and the gap and opposition between social groups deepen. 3. **Polarized content recommendation**: - **Definition**: In order to increase user engagement and click-through rate, algorithms tend to recommend more controversial, extreme or emotional content. - **Impact**: This type of content often exacerbates emotional reactions, increases users’ anger, fear or dissatisfaction, drives users to become more extreme, and strengthens social confrontation. 4. **Spread of false information and rumors**: - **Definition**: Algorithms may prioritize pushing content that is easy to attract users’ attention and share, including false information and rumors, because these contents are usually more attractive. - **Impact**: False information spreads rapidly among different groups, leading to misunderstandings, misjudgments and distrust, further dividing society. 5. **Information islands caused by content personalization**: - **Definition**: The information received by users is highly customized, lacking cross-group communication and understanding. - **Impact**: Different groups form their own information circles, lack a common cognitive basis, increase communication barriers, and intensify social divisions. 6. **Algorithmic bias**: - **Definition**: The algorithm design or training process may contain unconscious bias of developers, resulting in the priority display or marginalization of information from certain groups. - **Impact**: Specific groups may feel ignored or treated unfairly, increasing social dissatisfaction and division. 7. **Potential for manipulation and political propaganda**: - **Definition**: Criminals or political entities may use algorithms to promote specific political agendas or propaganda and influence public opinion. - **Impact**: Purposefully inciting group emotions, creating divisions and confrontations, and undermining social harmony. 8. **Short-term reward mechanism**: - **Definition**: Algorithms usually optimize short-term user interactions, such as clicks, shares, and likes, rather than long-term social health. - **Impact**: This leads platforms to prioritize content that can quickly attract attention, often sensational or extreme content, exacerbating social divisions. **Summary**: The algorithms of social media platforms have improved the user experience and the commercial value of the platform through mechanisms such as personalized recommendations, filter bubbles, and polarized content, but they have also invisibly deepened the gap and confrontation between various social groups. To alleviate this phenomenon, platforms need to pay more attention to information diversity and social responsibility when designing algorithms, and users also need to improve their information literacy and actively contact and understand different views.