Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
120 tokens/sec
GPT-4o
10 tokens/sec
Gemini 2.5 Pro Pro
44 tokens/sec
o3 Pro
5 tokens/sec
GPT-4.1 Pro
3 tokens/sec
DeepSeek R1 via Azure Pro
55 tokens/sec
2000 character limit reached

Beyond Likes: How Normative Feedback Complements Engagement Signals on Social Media (2505.09583v2)

Published 14 May 2025 in cs.HC

Abstract: Many online platforms incorporate engagement signals--such as likes and upvotes--into their content ranking systems and interface design. These signals are designed to boost user engagement. However, they can unintentionally elevate content that is less inclusive and may not support normatively desirable behavior. This issue becomes especially concerning when toxic content correlates strongly with popularity indicators such as likes and upvotes. In this study, we propose structured prosocial feedback as a complementary signal to likes and upvotes--one that highlights content quality based on normative criteria to help address the limitations of conventional engagement signals. We begin by designing and implementing a machine learning feedback system powered by a LLM, which evaluates user comments based on principles of positive psychology, such as individual well-being, constructive social media use, and character strengths. We then conduct a pre-registered user study to examine how existing peer-based and the new expert-based feedback interact to shape users' selection of comments in a social media setting. Results show that peer feedback increases conformity to popularity cues, while expert feedback shifts preferences toward normatively higher-quality content. Moreover, incorporating expert feedback alongside peer evaluations improves alignment with expert assessments and contributes to a less toxic community environment. This illustrates the added value of normative cues--such as expert scores generated by LLMs using psychological rubrics--and underscores the potential benefits of incorporating such signals into platform feedback systems to foster healthier online environments.

Summary

Integrating Normative Feedback with Social Media Engagement Signals

The paper "Beyond Likes: How Normative Feedback Complements Engagement Signals on Social Media" explores the limitations of engagement-driven metrics like likes and upvotes on social media platforms, noting that they can inadvertently amplify content that aligns with what's popular rather than what's constructive or inclusive. To address this, the authors develop a system for structured prosocial feedback using LLMs to evaluate content based on positive psychology principles. This research offers valuable insights into the potential of combining traditional popularity measures with normative cues to enhance the quality of online discourse.

Social media platforms have long been framed around engagement-centric designs, where likes, shares, and retweets are vital signals driving content visibility and user interaction. While effective for fostering user engagement, these metrics often favor emotionally charged or sensational content, leading to an echo chamber effect that can suppress more thoughtful or balanced discussions. The nexus of popularity and content quality presents a dichotomous landscape in which the unfettered freedom to promote any content can pose challenges to maintaining a civil and informed discourse online.

In examining these dynamics, the authors present a machine learning feedback system that scores content against indices drawn from positive psychology, such as the PERMA model, Digital Flourishing Scale, and VIA Character Strengths. The core hypothesis is that expert-based feedback can guide users towards more normatively aligned content even when it receives less peer endorsement. Their experimental design investigates the interaction between peer-driven and normative signals, focusing on user preferences when these signals diverge.

The findings demonstrate that expert feedback, when integrated with traditional engagement signals, can have a meaningful impact on guiding user behavior. In the absence of normative cues, users gravitate toward popular content. However, when expert feedback is available, there’s a noticeable shift towards higher-quality content, as evidenced by the increased selection of posts rated highly by the system's normative criteria. Specifically, when both peer and expert feedback were presented, users selected content with high normative scores 68.5% of the time, suggesting that expert-based evaluations can complement and counterbalance the effects of peer popularity signals.

From a theoretical standpoint, this dual feedback model situates itself within the proactive governance framework, advocating for an approach that not only discourages harmful content but proactively promotes positive contributions. The design implications are profound, as the integration of LLM-generated feedback offers a scalable mechanism for nudging users towards constructive behavior without imposing overly prescriptive or authoritarian measures.

Furthermore, this approach lends itself well to refining platform governance strategies. By blending peer and expert signals, platforms can create a more nuanced environment that respects both user autonomy and community standards. This method can be vital for platforms aiming to foster environments that are both engaging and enriching, amplifying valuable content that might otherwise be overlooked in the sea of sensationalism.

Future explorations could focus on refining the scoring mechanisms, perhaps by incorporating real-time user feedback on the efficacy of expert signals or examining how different content domains (e.g., political discourse, health misinformation) might benefit from tailored normative structures. Longitudinal studies could further assess the impact of such systems on user behavior over time, potentially informing the broader design and ethical considerations of AI-mediated content moderation.

In summary, the paper underscores the potential of integrating normative feedback systems with traditional engagement metrics to foster healthier online environments. By aligning digital spaces more closely with psychologically-grounded principles of well-being and prosociality, platforms can create interaction dynamics that support both user engagement and social value, enhancing the quality of discourse in our increasingly digital public spheres.

Dice Question Streamline Icon: https://streamlinehq.com

Follow-up Questions

We haven't generated follow-up questions for this paper yet.

Authors (2)