Personalised Yet Impersonal: Listeners' Experiences Of Algorithmic Curation On Music Streaming Services

Title of the Paper

Personalised But Impersonal: Listeners’ Experiences of Algorithmic Curation on Music Streaming Services

Bibliographic Information

• Subject Area: User experience and algorithmic recommendation (focused on music streaming platforms)
• Keywords: Music streaming, music interaction, algorithmic curation, music recommendation, vibe, personalization, user experience, algorithmic control, music memory, social listening

Research Background and Issues

• Identified Problems or Challenges:
  Algorithmic recommendation and curation features widely used in music streaming services have profoundly impacted user behavior. However, despite the convenience offered by these algorithms, many users perceive these features as lacking personalization, failing to meet specific needs, and occasionally feeling "cold" or "misaligned."

• Significance:
  Music plays a vital role in daily life, serving core functions such as emotional regulation and identity expression. Music streaming services aim to optimize user experience through personalized recommendations. However, this optimization may introduce potential issues related to transparency, bias, and user control.

• Research Motivation and Related Work:

  • The study seeks to understand how users perceive algorithm-based music recommendation systems (e.g., Spotify, Apple Music) and focuses on the impact of these features on listening habits and experiences.
  • It references existing studies, such as Liikkanen’s research on music interaction paradigms (Curated mode and On-Demand mode) and issues of transparency and user control in recommendation systems, aiming to update and expand the understanding of music streaming experiences.

Proposed Solution

• Proposed Approach or Research Methodology:

  • The authors adopt "Algorithmic Experience" (AX) as an analytical framework, focusing on user interaction experiences with algorithmic recommendations.
  • The study consists of two phases: the first involves user interviews and observations, while the second employs design workshops to invite user participation and further explore their needs and ideas.

• Innovative Aspects:

  • Introduces the concept of "vibe" to describe recommendations that need to align not only with musical content similarity but also with users' current emotions and contexts, making recommendations more "personalized."
  • Explores the tension between "personalized and impersonal" in music recommendation systems and how design can achieve greater transparency and user control.

• Implementation Steps and Methods:

  1. Interviews and Observations: Conducted interviews with 15 Spotify and Apple Music users aged 18-36 to explore their listening habits and experiences with recommendation features.
  2. Participatory Design Workshops: Used design tasks and dialogue templates to invite users to design their ideal music recommendation system features, focusing on personalization, flexible control, and transparency.
  3. Data Analysis: Applied Braun and Clarke’s thematic analysis method to systematically categorize and summarize the data collected from interviews and design workshops.

Research Findings

• Key Findings:

  1. Three Listening Modes: Users have different needs and experiences in three listening scenarios—Background mode, Curation mode, and Discovery mode.
  2. Concept of "Vibe": Proposed as a critical criterion for evaluating whether recommended content matches users’ current context and inner feelings. "Vibe" serves as a core metric for users to assess music recommendations, addressing the limitations of overly rigid technical metrics.
  3. Tension in User Needs: Users desire more personalized control (e.g., adjusting style and diversity via "sliders") but also wish to reduce daily interaction pressure (e.g., typical Lean-back mode).

• Experiments and Evaluations:

  • Most users expressed a desire for greater control over personalized recommendations, not only regarding content diversity but also in the operation and presentation of recommendation algorithms (e.g., transparency or virtual personas).
  • User designs proposed innovative solutions combining "transparent control" (e.g., visibility of recommendation sources) with "entertainment interaction" (e.g., customizable virtual recommender appearances).

• Comparison with Existing Solutions:

  • Current music streaming services lack significant user control mechanisms. This study emphasizes control while introducing "vibe" and "mnemonics" (memory recall) as key factors influencing the design of recommendation algorithms.

• Limitations and Future Directions:

  1. Study Sample: The small sample size (15 participants) and focus on young users limit the ability to comprehensively address the diverse needs of other demographic groups.
  2. Future Work:
     • Investigate the influence of variables such as music interests, age, and listening frequency on interaction mode preferences.
     • Expand comparative studies across different platforms.
     • Explore designers’ perspectives to optimize recommendation algorithms based on user feedback.

Conclusion

This study highlights the core tension between "personalization" and "user control" in music streaming services, demonstrating that users expect platforms to provide more flexible and transparent algorithmic recommendations that align with their emotions, contexts, and intentions. Furthermore, the concept of "vibe" offers a novel perspective beyond traditional algorithmic metrics, urging academia and industry to develop more user-centered and personalized music recommendation experiences in the future.