Group Polarization / 群体极化

Definition

Group polarization is the intensification of positions or intergroup hostility when identity, repeated in-group signals, local norms, platform structure, and emotionally adversarial contact push people away from moderation.

Current Synthesis

The bounded evidence separates informational, social, algorithmic, and affective routes. Shared leaning can intensify when stronger in-group expressions reset the local center and moderation becomes costly to status or belonging. Algorithms can reinforce this process by repeatedly presenting inferred peers and identity categories, but the evidence does not support a universal sealed filter bubble: multi-platform use and incidental exposure can broaden contact.

Exposure alone is not correction. Opposing views delivered through rage-rich feeds, angry comments, adversarial formats, mass reporting, and forced position-taking can broaden information while worsening Affective Polarization / 情感极化. EP113’s Xiao Zhan/AO3 case adds reciprocal escalation: identity defense can organize punitive action, while opponents can answer with pressure on the idol, fandom, and brands. The Eagleman interview adds an older and lower-level identity layer: religious, political, sporting, or even arbitrary labels can alter empathic response, while dehumanizing propaganda and zero-sum models can make harm easier to justify. Social media may amplify visibility and conflict without having invented polarization.

Key Claims

• Repeated in-group argument can move the perceived group center toward a more extreme position.
• Loyalty, safety, and social status can make moderation or belief revision costly.
• Algorithmic prediction can turn inferred similarity into repeated exposure and social self-recognition.
• Cross-viewpoint exposure can increase hostility when the surrounding format rewards conflict.
• Polarization predates social media, though platforms can amplify, sort, and make it more visible.
• Minimal group labels can affect empathic response without determining later conduct.
• Dehumanizing language, reporting, brand pressure, and forced position-taking can convert identity difference into reciprocal group punishment.

Evidence

• Norm intensification - 132.争论与说服：我们为什么吵架，怎么才算赢？ links group polarization to identity, tribal truth, information cocoons, and stronger local expressions.
• Calculated publics - 159.算法的六副面孔：它是如何从处理数据，变成定义我们是谁的 describes algorithms repeatedly feeding users inferred groups, tastes, and worldviews.
• Affective qualification - 164.算法的“兔子洞”：为什么你总在看完新闻后滑向娱乐？｜对谈黄圣淳教授 says opposing exposure can intensify hostility inside conflict-rich platform formats.
• Historical and identity boundary - Science & Tools of Learning & Memory | Dr. David Eagleman argues that polarization predates social media and describes empathy differences across religious, atheist, sport, and arbitrary groups.
• Dehumanization route - Science & Tools of Learning & Memory | Dr. David Eagleman connects propaganda labels, reduced person-modeling, and zero-sum thinking to intergroup harm.
• Reciprocal mobilization - EP113-充满偏见的互联网时代，我们更接近真相了嘛？ uses the Xiao Zhan/AO3 controversy to connect Fandom Collective Action / 饭圈集体行动, Weaponized Reporting / 举报武器化, label-driven camp assignment, counter-pressure, and silence under threat.

Counterevidence & Qualifications

The sources do not show that all homogeneous groups polarize, that algorithmic feeds universally seal users into cocoons, that exposure to disagreement is always harmful, or that an fMRI contrast predicts individual behavior. EP113 does not provide fan demographics, complaint records, a platform audit, or independent reconstruction of the AO3 access change. Political conflict also depends on institutions, leaders, material stakes, history, and power. Social media can amplify and reveal division without being its sole cause.

What Changed

• Migrated the page to synthesis-v1 while preserving the original evidence order.
• Added polarization’s pre-social-media history and separated platform amplification from origin.
• Added identity-gated empathy, dehumanizing labels, and zero-sum models as qualified mechanisms.
• Added reciprocal fandom escalation, punitive reporting, and silence pressure as source-scoped mechanisms.

Related Concepts

• Information Cocoon / 信息茧房 - media setting that can narrow repeated exposure without being universally sealed.
• Tribal Truth / 部落真相 - belonging mechanism that makes belief revision socially costly.
• Calculated Publics / 计算出的公众 - algorithmically inferred groups that feed identity back to users.
• Affective Polarization / 情感极化 - emotional hostility that can increase despite informational contact.
• Group-Identity Empathy Gating - lower-level identity response that does not determine conduct.
• Dehumanization By Comparison / 比较导致的去人性化 - loss of concrete personhood that can make group harm easier to justify.
• Constructive Dissent / 建设性异议 - corrective practice for widening evidence without demanding identity surrender.
• Fandom Collective Action / 饭圈集体行动 - coordination layer that can translate polarized identity into data, consumption, defense, and pressure.
• Weaponized Reporting / 举报武器化 - enforcement channel that can convert factional conflict into material exclusion.