Strategy brief: Collective Judgment and Digital Witch Hunts in the Age of Social Media
The Status Quo
The Modern Information Environment
Since the emergence of digital platforms, social media has fundamentally transformed the way humans live: how we connect, communicate, consume information, and understand events happening around us. If, in the past, communication often depended on slower interpersonal, face-to-face interactions or traditional media systems, today, people can encounter strangers, narratives, rumors, and public controversies from and across the world within seconds. This transformation has created a new information environment where information does not simply move from one person to another; it circulates rapidly through networks shaped by digital platform affordances, algorithmic ranking systems, engagement metrics, and user participation.
This unprecedented speed of information circulation and scale of participation can be incredibly powerful. Digital platforms allow ordinary users to engage in public discourse, organize around shared concerns, mobilize communities, and bring visibility to issues that may otherwise remain ignored. At the same time, however, the very features that make digital platforms effective tools for activism and collective action can also make them powerful catalysts for rumor, speculation, and premature collective judgment. The challenge facing today’s information environment is not simply that false information spreads online; rather, narratives often gain visibility, legitimacy, and influence before facts have been fully established. In many cases, people are not reacting to verified information itself, but to interpretations, assumptions, and emotionally compelling narratives that emerge by creators online, while important details remain uncertain.
In digital media environments, it is important to recognize that the voices that become most visible online today are not necessarily the most representative, credible, or well-informed ones. Unlike traditional information systems, where information is often passed through institutional gatekeepers such as journalists, editors, experts, or public officials, today’s social media platforms allow virtually anyone to participate in public discourse and contribute to the circulation of information. Us too, can become a content-creator and create, remix, and reproduce others’ content. While this democratization of participation can be empowering, it also means that visibility is often determined by engagement rather than accuracy. Platform affordances such as algorithmic recommendation systems, reposting features, comment sections, trending pages, and engagement-based ranking mechanisms can amplify content that attracts attention, regardless of whether it is accurate, representative, or sufficiently verified.
At its core, the information integrity threat examined in this brief is not simply the circulation of false information, but the growing tendency for collective judgments to form before facts and context have been sufficiently verified. These unverified truths are becoming the foreground of people’s judgements and opinions, and are leaving some extremely hurt. Unlike traditional misinformation concerns, which focus primarily on whether information is true or false, this threat emerges even when facts remain uncertain. The problem is not only that incorrect narratives spread, but that public judgments and real-world consequences often emerge before sufficient evidence is available to justify them.
The Challenge in Context
Across different cultural, political, and social contexts, digital platforms have repeatedly demonstrated how narratives can gain legitimacy and influence before facts and context have fully stabilized. The following cases illustrate not only the consequences of premature collective judgment, but also why addressing this issue is particularly difficult for platform Trust & Safety teams.
The case of South Korean actress Kim Sae-ron demonstrates the human consequences of this dynamic. Following a 2022 drunk-driving incident, Kim faced years of criticism across social media, entertainment news outlets, and celebrity gossip channels. According to reporting by The Guardian, public scrutiny gradually expanded beyond the original incident itself, with users criticizing her private life, questioning the sincerity of her remorse, and attacking her attempts to rebuild her career (Rashid & McCurry, 2025). What began as public criticism of a specific action evolved into a broader narrative about her character, morality, and worthiness of forgiveness. As these narratives circulated repeatedly across digital platforms, visibility increasingly became a source of legitimacy. The more frequently users encountered criticism and speculation, the more socially accepted and unquestioned those narratives appeared, and the more it hurt the actress, cornering her to death. This case illustrates how digital environments can transform accountability into something far more expansive, allowing social consequences to accumulate long before a complete understanding of a person’s circumstances has room to emerge.
A similar pattern can be observed in the aftermath of Indian actor Sushant Singh Rajput’s death in 2020. However, unlike the Kim Sae-ron case, the central issue was not the persistence of judgment but the role of uncertainty itself. As Banerjee (2020) describes, public discourse rapidly expanded beyond available evidence and became intertwined with speculation, conspiracy theories, and what he terms a “speculative witch-hunt.” Traditional media outlets, influencers, online communities, and ordinary social media users collectively participated in constructing explanations for what had happened and who might be responsible. This case is particularly significant because it demonstrates that digital witch hunts do not always emerge from clearly false information, but often emerge from informational gaps. When facts remain incomplete, people naturally attempt to construct explanations, connect fragments of evidence, assign motives, and fill uncertainty with interpretations that feel plausible. Through repeated sharing, discussion, and engagement, those interpretations can gradually begin functioning as evidence themselves. In this way, uncertainty does not necessarily slow collective judgment; under the right conditions, it can accelerate it.
At the same time, digital amplification itself is not inherently harmful. Kang’s analysis of South Korea’s 2008 Mad Cow protests demonstrates how the very same affordances that contribute to rumor circulation and speculative judgment can also facilitate civic engagement, information sharing, and democratic participation (Kang, 2017). Through blogs, online communities, and discussion forums, citizens shared concerns regarding government decisions on U.S. beef imports and mobilized large-scale candlelight movements. Digital networks enabled information to circulate rapidly, helping ordinary citizens participate in public discourse and collective action on a scale that would have been difficult through traditional communication channels alone. This case serves as an important reminder that amplification is not the problem in itself, but the core challenge lies in fostering healthier information environments that preserve amplification’s benefits while reducing the likelihood that visibility becomes mistaken for credibility and that speculation is treated as established fact.
Strategic Recommendations for Platform Trust & Safety Teams
One: Introduce Greater Friction Around Rapidly Spreading Accusations
One of the defining characteristics of digital witch hunts is the speed at which accusations, interpretations, and judgments can spread across networks. Once a narrative begins gaining traction, users can repost, comment on, and amplify claims within seconds, often without engaging with additional context or verified information.
To address this problem, platforms should introduce greater friction around the circulation of rapidly spreading accusations involving identifiable individuals. In this context, friction refers to platform design interventions that intentionally slow the momentum through which accusations, interpretations, and judgments travel across networks. Much like friction in physical systems reduces speed and momentum, platform friction can create moments of pause and reflection before users repost, comment on, or amplify potentially harmful narratives. The goal is not to stop discussion altogether, but to reduce the likelihood that collective judgment forms before sufficient evidence is available.
Possible interventions include prompts encouraging users to review additional context before reposting, temporary warnings attached to highly viral accusation-based content, or brief delays before users can repeatedly share the same claim. Platforms could also introduce uncertainty labels on content associated with developing controversies, such as “Evidence Still Emerging,”“Developing Story,” or “Claims Unverified.” For example, TikTok could attach such notices to rapidly trending videos discussing unresolved allegations and require users to actively click through before viewing the content, creating a brief moment of reflection while reducing passive exposure during scrolling on for-you-page. Instagram could display uncertainty labels on highly shared controversy-related posts and Stories, directing users to verified updates before users engage further with the content. Similarly, Reddit moderators could receive automated prompts encouraging the creation of pinned discussion threads that consolidate verified updates and reduce rumor fragmentation across multiple posts, while X could attach visibility notices to rapidly spreading accusations that remain unresolved.
Given MIT’s finding that false news on Twitter spread farther, faster, deeper, and more broadly than truthful information because users were drawn to its novelty and emotional appeal (Dizikes, 2018), even small increases in friction may help reduce the speed at which unverified narratives gain momentum. More importantly, such interventions would help preserve a space between discussion and judgment, allowing facts, context, and verification processes greater opportunity to catch up before narratives solidify into socially accepted truth.
Two: Reduce the Algorithmic Amplification of Unverified Narratives
Previous researches have revealed that visibility on digital platforms is not necessarily a reflection of accuracy, representativeness, or consensus. Kim and colleagues found that highly engaged users and toxic comments often receive disproportionate visibility online, causing extreme viewpoints to appear more common than they actually are (Kim et al., 2021). Similarly, Piccardi and colleagues found that recommendation systems can shape users’ attitudes and perceptions rather than merely reflect them (Piccardi et al., 2025). Together, these findings suggest that platform algorithms play an active role in determining which narratives gain prominence within public discourse.
Acknowledging this challenge, platforms should consider adjusting recommendation systems during rapidly developing controversies involving identifiable individuals. Instead of relying primarily on engagement metrics such as likes, shares, comments, watch time, or reposts when determining visibility, platforms could incorporate signals related to source credibility, evidentiary support, and informational diversity. For example, TikTok and Instagram could temporarily reduce the recommendation weight of accusation-based content experiencing unusually rapid growth, while Reddit and X could prioritize posts linking to original reporting, verified information, or reputable sources. Platforms could also consider reducing the visibility of engagement metrics such as disabling on displaying the like counts, repost counts, or upvote totals on highly contentious and developing stories, limiting the tendency for users to interpret popularity as evidence of accuracy or consensus.
The goal is not to suppress discussion or determine which viewpoints are correct, but to reduce the likelihood that engagement alone becomes mistaken for credibility. By limiting the algorithmic amplification of highly speculative content during periods of uncertainty, platforms can create healthier information environments where narratives compete on the basis of evidence and context rather than virality alone.
Three: Increase the Visibility of Corrections, Updates, and Context
Digital platforms often excel at distributing accusations, outrage, and emerging narratives but perform far less effectively when distributing corrections, clarifications, and updated information. Once a narrative becomes established, subsequent evidence frequently receives significantly less attention than the original claims that first captured public interest. As a result, users may continue operating with outdated understandings of an event even after important new information becomes available. Across all three cases discussed, no matter the outcomes, one similarity they shared was that these were all developingnews. If platforms continue prioritizing initial engagement while neglecting later corrections, users may never encounter the information necessary to revise their understanding.
Taking this into consideration, platforms should invest in systems that actively elevate corrections, contextual updates, and newly verified information related to highly visible controversies. Users who previously engaged with viral accusations could be shown verified updates as developments emerge, while recommendation systems could prioritize authoritative follow-up reporting when users continue searching for or interacting with ongoing controversies. One particularly promising intervention would be the creation of platform-generated update panels that aggregate the latest developments from trusted and established news organizations. Beyond simply determining what is “true,” these panels would shift the focal focus and provide users with the most recent verified reporting available at a given moment. Similar to election-information centers or crisis-information hubs already used on some platforms, these updates could appear above comment sections, alongside search results, or attached to highly viral posts, given that many tend to click the comment section in viral posts.
If platform design helps determine which narratives become prominent, platforms also have a responsibility to ensure that corrections and evolving context receive meaningful visibility. By making updates and corrections more accessible throughout the lifecycle of a story, platforms can create healthier information environments that encourage ongoing learning and revision rather than premature certainty, keeping the users “up-to-date”.
Conclusion
The challenge examined in this brief is not simply the spread of false information, but the speed at which narratives become socially meaningful before facts fully become stabilized. In today’s digital information environment, visibility can easily become mistaken for credibility, encouraging individuals to form judgments and participate in collective narratives while important uncertainties remain unresolved. Yet the same systems that contribute to these risks also enable civic engagement, public accountability, and collective action. The challenge, therefore, is not choosing between participation and protection, but designing platforms that make space for both. Together, the recommendations proposed in this brief offer a three-layered approach that slows premature amplification, reduces the visibility advantages of unverified narratives, and strengthens the circulation of updated information. In doing so, platforms can help foster healthier information environments where public understanding has greater opportunity to evolve alongside evidence rather than ahead of it.
References
Kim, J. W., Guess, A., Nyhan, B., & Reifler, J. (2021). The distorting prism of social media: How self-selection and exposure to incivility fuel online comment toxicity. Journal of Communication, 71(6), 922–946. https://doi.org/10.1093/joc/jqab034
Piccardi, T., Saveski, M., Jia, C., Hancock, J., Tsai, J. L., & Bernstein, M. S. (2025). Reranking partisan animosity in algorithmic social media feeds alters affective polarization. Science, 390(6776). https://doi.org/10.1126/science.adu5584
Kharazian, Z., & Hill, B. M. (2026). Epistemic Infrastructure under Attack: Conceptualizing the Capture and Destruction of Public Knowledge Institutions. https://doi.org/10.31235/osf.io/vxumb_v1
Banerjee, D. (2020). ‘All that followed a death. . .’: An alleged celebrity suicide, media discourse and mental health. International Journal of Social Psychiatry. https://doi.org/10.1177/0020764020985572
Kang, J. (2017). Internet activism transforming street politics: South Korea’s 2008 ‘mad cow’ protests and New Democratic sensibilities. Media, Culture & Society, 39(5), 750–761. https://doi.org/10.1177/0163443717709444
Dizikes, P. (2018, March 8). Study: On Twitter, false news travels faster than true stories. MIT News | Massachusetts Institute of Technology. https://news.mit.edu/2018/study-twitter-false-news-travels-faster-true-stories-0308
Rashid, R., & McCurry, J. (2025, February 22). “like a giant Squid Game”: Soul searching in South Korea after latest celebrity suicide. The Guardian. https://www.theguardian.com/world/2025/feb/22/like-a-giant-squid-game-soul-searching-in-south-korea-after-latest-celebrity-suicide
