The "Recommendations" list is a popular engagement feature used in various types of apps, from music streaming to news outlets. Recommendations enhance user satisfaction, prolong sessions, and increase retention rates. Although the pattern appears similar across all contexts and the purpose is clear ("Give me more of this"), a one-size-fits-all algorithm can lead to different outcomes.
Let's go by example.
Next song
Consider a music streaming service. Users are happy when the app continuously plays songs with similar acoustic characteristics. This is because they likely want to maintain their current mood. Listening to similar songs minimizes deviations in mood. Let's refer to this task as "finding other similar units."
Next article
Now think about a news aggregator employing the same algorithm. Users are presented with articles that share similar ideas. Reading articles with similar ideas reinforces existing opinions, leading to a resistance to alternative viewpoints. It forms a "bubble." What perhaps could be better is to "find other units on a similar topic." This would help users develop a more nuanced understanding of the topic and foster resistance to manipulation.
But hey, things that make us angry keep us more engaged.