Music box that is. For those of you who have not seen it, Pandorais a music finding service developed by the Music Genome Project. This is a fascinating site and uses some fairly spiffy analytics to find music like the music you like. Both my partner and I have played with it and had a mixed experience. You can say "I like this" or "I don't like this" and the software will tell you why it selected a particular song, for example:
"we're playing this track because it features a subtle use of vocal harmony, mild rhythmic syncopation, mixed acoustic and electric instrumentation, major key tonality and melodic songwriting"
Now this is all very clever and it uses multiple song likes and dislikes to refine it's analysis. When my partner was using it she identified what I think is the problem - there is no way to apply rules on top of the analytics.It does not let you say "Yes, but I like this song DESPITE the syncopation" or "Stop recommending female vocalists" - you must rely only on tuning the analytic.As such, it makes it harder to refine your preferences than seems reasonable. It is the power ofcombining rules (preferences being one example of rules) with analytics that makes Enterprise Decision Management so powerful. Rules can often be great and analytics, like those in Pandora, very clever. Combining them will almost always result in a better decision.