Reframing the filter bubble through diverse scale effects in online music consumption

Area of Science:

• Information Science
• Human-Computer Interaction
• Computational Social Science

Background:

• The impact of algorithmic curation on user consumption diversity is a widely debated topic.
• Existing research often compares algorithmic recommendations to organic user behavior.
• Contradictory findings suggest scale and information representation influence outcomes.

Purpose of the Study:

• To empirically analyze the impact of algorithmic curation on user consumption diversity.
• To investigate how different scales and platform affordances affect diversity metrics.
• To provide a nuanced understanding of algorithmic influence beyond the filter bubble concept.

Main Methods:

• Analysis of approximately 50,000 Deezer users' artist consumption histories.
• Examination across three scales: intra-sessions, inter-sessions, and inter-affordances.
• Comparison of consumption diversity under varying degrees of algorithmic assistance.

Main Results:

• Algorithmic curation may introduce more novelty compared to organic user behavior.
• The novelty introduced by algorithms tends to be more semantically confined.
• Different scales and information representations yield contrasting conclusions about diversity.

Conclusions:

• Algorithmic curation's effect on diversity is complex and depends on perspective.
• The filter bubble narrative needs updating to account for confined novelty.
• Understanding these dynamics is crucial for on-platform algorithmic systems.