NIPL/Assumptions

NIPL is based on the following assumptions, which are yet to be actually verified... this kind of needs to be done...

• Songs can be rated on a scale from bad to good
  • This is fairly accepted in fact, and is a basis for any song rating system

• Good human-generated songlists are based not on a simple rating of the songs, but just as (or more) importantly, on the flow of songs, from one to the next.
  • This is pretty easily accepted. It's obvious that Counting Crows will follow Crowded House alot better than, say, Monty Python.

• The progression from one song to the next can be rated on a scale from bad to good.
  • This is also fairly easily accepted.

• The score of songA->songB can be used as an approximation for artistA->artistB (and similarly for albumA->albumB and genreA->genreB.
  • This seems to make sense. For example, given that Smells Like Teen Spirit doesn't really work well as the song after Orinoco Flow, this is an indication that Nirvana should never follow Enya, that grunge should never follow melodic folk pop, and so on.

• The score of songA->songB is always likely to generate a similar value on subsequent rankings.
  • This is hardest to justify, since a change in song is likely to be a change in mood and genre, no matter how slight. If the change in mood heralded by the new song is appropriate to your current listening experience, then it will be a good change. On the other hand, it might clash with your mood...
  • for eg: Billy joel -> Bob Dylan = good if you want to listen to more folkey/lyricy stuff next, but = bad if you want to progress to more rocking stuff.
  • This scoring-according-to-context probably needs to be taken into account in NIPL, but may add another layer of complexity onto the system - which is already overly complex (some might say)