Designing Usable Controls for Customizable Social Media Feeds

2025-09-23

AI Summary

Current social media recommendation algorithms lack user controllability, undermining user autonomy and value alignment; existing algorithmic governance mechanisms remain underutilized due to poor usability. This paper proposes a deeply integrated, intuitive, and expressive algorithmic control mechanism embedded directly within the information feed. It innovatively introduces *teachable feedback*—a lightweight, in-situ interaction paradigm that enables real-time, low-friction user input during routine browsing, thereby unifying algorithmic regulation with natural user behavior. Implemented on the Bluesky platform as the Pilot system, the design incorporates explainable real-time feedback and human-AI co-design principles. A user study demonstrates statistically significant improvements in perceived agency (p < 0.01), enhanced comprehension of algorithmic logic, and increased metacognitive reflection and intentional adjustment of content consumption habits.

Technology Category

• Humans and AI: Intelligent User Interfaces
• Search and Optimization: Algorithm Configuration
• Machine Learning: Learning Preferences or Rankings

Application Category

• Responsible Web: Human-perceived consequences of algorithmic deployment on the web
• User Modeling, Personalization and Recommendation: Metrics for user behavior and evaluating success
• Economics, Online Markets and Human Computation: Economics and fairness of platforms and recommendation systems

Abstract

Personalized recommendation algorithms deliver content to the user on most major social media platforms. While these algorithms are crucial for helping users find relevant content, users lack meaningful control over them. This reduces users' sense of agency and their ability to adapt social media feeds to their own needs and values. Efforts have been made to give users more control over their feeds, but usability remains a major barrier to adoption. Drawing upon prior work in designing teachable social media feeds, we built Pilot, a novel system of controls and feedback mechanisms on BlueSky that are expressive, intuitive, and integrated directly into the feed to allow users to customize their feed while they browse. Our user study suggests the system increases the user's sense of agency, and encourages them to think more critically about curating their feeds. We synthesize design implications for enhancing user agency over social media feeds.

Problem

• Users lack meaningful control over social media recommendation algorithms
• Existing feed controls face usability barriers that limit adoption
• Current systems reduce user agency and feed customization ability

Innovation

• Expressive intuitive controls integrated into feed
• Novel system with feedback mechanisms on BlueSky
• Enables real-time feed customization while browsing