This way of predicting tastes by aggregating millions of people’s preferences is known as “collaborative filtering”—collaborative because it takes many users’ inputs, and filtering because it uses the data to narrow down the next thing you want to hear.