Nick Seaver discussing his book, Computing Taste: Algorithms and the Makers of Music Recommendation

Summary

In this insightful interview, Professor Siever discusses the concept of 'attention' in the digital age, particularly in relation to AI and machine learning. He explores how systems like recommendation algorithms and generative models (e.g., ChatGPT) manipulate attention, shaping user behavior and influencing the attention economy. Siever examines the implications of attention overload, including the rise of generative content and its potential to overwhelm users. His current research, focusing on 'attention fragments', delves into the diverse ways attention is understood across fields like psychology, virtual reality, and web development, offering a critical perspective on the societal consequences of attention-driven technologies.

Takeaways

• 😀 **Computing Taste**: Nick Siever's book explores how music recommendation systems shape cultural products and influence user behavior through algorithms.
• 😀 **Anthropological Perspective**: Siever uses anthropology to understand recommendation systems, focusing on the social aspects of technology rather than just the technicalities.
• 😀 **Technical vs Folk Language**: The book highlights the importance of both technical and folk vocabularies in understanding how engineers conceptualize and explain their work.
• 😀 **Captology**: Siever discusses *captology* (persuasive computing), exploring how recommendation systems use persuasive techniques to capture attention and influence user behavior.
• 😀 **Spatial Metaphors in Data**: Engineers often use spatial metaphors (like 'gardens' or 'parks') to describe their work in data management, reflecting a view of themselves as caretakers of data ecosystems.
• 😀 **Attention Economy**: The concept of the *attention economy* is central to the book, with Siever examining how algorithms are designed to capture and monetize users' attention.
• 😀 **Attention as a Commodity**: Siever links the commodification of attention to concerns about mental overload, distraction, and the ethical issues surrounding algorithmic manipulation of user focus.
• 😀 **Denial of Attention Attacks**: Siever references a concept from sociologist Zainab Tufati called *denial of attention attacks*, describing how overwhelming content generation (by AI) can disrupt users' attention and decision-making.
• 😀 **Generative AI and Attention**: The rise of generative AI systems like ChatGPT and DALL·E raises questions about the flood of machine-generated content and its impact on human attention and creativity.
• 😀 **Future Research on Attention**: Siever's ongoing research, titled *Attention Fragments*, will explore how attention is understood and managed across different fields, such as psychology, web development, and virtual reality.

Q & A

What inspired Professor Nick Siever to write his book 'Computing Taste: Algorithms and the Makers of Music Recommendation'?

Professor Siever's inspiration for the book came from his background in media studies and a fascination with the intersection of technology and culture, especially in music. He had previously worked on the history of the player piano, which led him to explore music as a domain where technology and cultural expression collide. His interest in ethnographically studying music recommendation systems grew from there.

How does Professor Siever view music recommendation systems from an anthropological perspective?

Professor Siever views music recommendation systems as a rich site for anthropological study because they exemplify the tensions between technology and culture. While music is often seen as inherently human and expressive, it is also deeply technical, involving various machines and algorithms. He uses ethnography to explore how engineers involved in these systems think about music, their goals, and how they sometimes work with assumptions about human behavior.

What is captology and how does it relate to music recommendation systems?

Captology, coined by BJ Fogg, refers to the study of computers as persuasive technologies. It deals with how technologies, including music recommendation systems, are designed to persuade or influence users' behavior. Siever applies this concept to music recommendation systems, suggesting that these systems are built not just to recommend music but to capture user attention and influence their desires, often using techniques that tap into psychological models of behavior.

How does Professor Siever explain the role of spatial metaphors in music recommendation systems?

Siever discusses how spatial metaphors are used in music recommendation systems to map user preferences within similarity matrices. These systems often treat musical genres and tastes as 'spaces' where similar music is grouped together, and users navigate this space to discover new content. These spatial metaphors help engineers understand and present data in a way that feels both scientific and intuitive, even though it oversimplifies the complexity of cultural and social dynamics.

What critiques does Professor Siever offer about the use of spatial metaphors in algorithmic systems?

Siever critiques the naturalizing use of spatial metaphors in algorithmic systems, especially in machine learning. These metaphors present data as if it naturally exists in space, which can obscure the social and cultural work involved in data creation. He compares these metaphors to earlier forms of computer programming, suggesting that by likening machine learning to farming or gardening, engineers might be attempting to distance their work from earlier, more deterministic forms of programming.

What does Siever mean by 'denial of attention attacks' in the context of generative AI systems?

Siever refers to Zainab Tufati's concept of 'denial of attention attacks,' describing how generative AI systems like ChatGPT can overwhelm people with machine-generated content, making it harder for individuals to focus and discern meaningful information. This analogy compares it to a denial of service attack in computing, where a system is flooded with excessive requests, leading to user distraction or disengagement.

What role does attention play in the design of music recommendation systems?

Attention is a crucial metric in music recommendation systems, often used as a proxy for user satisfaction. Metrics like dwell time, which tracks how long a user engages with content, are used to measure attention. However, Siever notes that attention is a complex and multifaceted concept that isn't always accurately captured by such simple metrics, as users might engage with content they are dissatisfied with or may be passively exposed to.

What is the significance of 'dwell time' as a measure of attention in recommendation systems?

Dwell time is a commonly used metric in recommendation systems that measures how long a user spends interacting with a piece of content, such as a song or video. It serves as a rough proxy for user satisfaction, under the assumption that the longer a user engages with content, the more likely they are to be satisfied. However, Siever acknowledges that this isn't always the case, as users might engage with content out of habit or for reasons unrelated to satisfaction.

How does Siever critique the attention economy in the digital age?

Siever critiques the attention economy by pointing out that attention has become a commodified resource in digital platforms, especially in social media and streaming services. While attention is often seen as a key to success for these platforms, Siever suggests that the focus on capturing attention may overlook the complexities of how users experience and engage with content. The obsession with attention metrics might also lead to the exploitation of users' psychological vulnerabilities.

What is the focus of Professor Siever's current research project on attention?

Siever's current research project, tentatively called 'Attention Fragments,' focuses on understanding attention as a cultural concept. He is exploring how different fields—such as psychology, web development, virtual reality, and facial recognition—conceptualize and use the idea of attention. His project examines how attention is mobilized in various domains, drawing on interviews with professionals in these fields to understand the broader implications of attention in technology.