Popularity measures how much attention or use content attracts across a population. Recommender systems may treat it as a practical criterion for distributing content, especially when abstract user value cannot be measured. section 1.3.3
Popularity is not the same as quality or value. Engagement can act as a rough behavioural signal, but clickbait shows why attention does not reliably indicate usefulness. Popularity can also reproduce itself when prominent recommendations receive more attention and become even more popular. section 1.3.3
A streaming-app feedback loop
A music app displays an artist’s most-listened-to songs. Users choose them because they are prominent and save browsing time, increasing their listening counts and reinforcing their popularity. section 1.3.3
Popularity therefore reflects an interaction among users, algorithms, and other users: recommendations influence attention, while user behaviour changes what algorithms promote. section 1.3.3

