I’m particularly interested in the skip degradation of a given song’s score. The reason for this is that it would be interesting to see if partial skips (skips at a later time point in the song) are taken into account and whether the weight of such a “late skip” is lower than an “immediate or early skip”.

In the past, I’ve been annoyed by many shuffle functions in programs and played around with pattern recognition as a way to determine the current “listening mood” of the listener and weight songs by that. In such a function, you also have to go by song skipping to determine a song’s weight. (By the way, it was mostly an exercise in neural network programming, not a “real” development project.)