Efficient Data Retrieval and Comparative Bias Analysis of Recommendation Algorithms for YouTube Shorts and Long-Form Videos

Short-form videos promote faster content narrowing: The study found that YouTube Shorts' recommendation algorithm shifts users toward engaging but less diverse content more quickly than long-form video recommendations, potentially limiting exposure to varied perspectives in shorter timeframes.

Recommendation algorithms can amplify specific narratives on sensitive topics: The research identified biases in how algorithms recommend content about politically sensitive issues like the South China Sea dispute, demonstrating that these systems can shape which viewpoints users encounter rather than presenting balanced information.

Transparency requires independent research methods: Because YouTube's official API has significant limitations, researchers had to develop alternative data collection frameworks using parallel computing and advanced scraping techniques—highlighting that external oversight of recommendation systems may require workarounds to assess algorithmic fairness.

Different video formats warrant different oversight approaches: The distinct behavioral patterns between short-form and long-form video recommendations suggest that responsible AI governance cannot use a one-size-fits-all approach; platforms need format-specific strategies to address bias and diversity concerns.

Echo chamber risks vary by content type: The findings suggest short-form video platforms may be particularly susceptible to creating echo chambers due to faster algorithmic narrowing, indicating this format may require additional safeguards to maintain content diversity and user exposure to opposing viewpoints.