There are many well designed playlist algorithms. I am a repeater myself meaning, whenever i fall in love with a song i repeat it very, very often and usually for a long time. As I know several other people with the same behaviour I thought of a new playlist that acts like "mix of the week" and gives a collection of songs that have been listened by other repeater-listeners very often and has simliarities to the person.
What is the difference? Not only how often it has been hear, but also how often in a row - meaning for people that like to do this it may be a predicting variable whether the next repeater may love that song too.