My Soundblaster digital music system doesn't have the "shuffle" feature, but instead it has these "Smart Playlists" that categorize music according to characteristics like when a track was added or when it was last played. So I can choose to play "most frequently played" or "newest" or, as I've taken to doing a lot, "old favorites." These are the songs you play the least. Some of them I don't even recognize. But others are forgotten gems. As with any "shuffle" function there are abrupt transitions when techno gives way to classical to bluegrass. This morning we've heard Underworld, then Bob Dylan, and now the Police are singing to me while I eat my oatmeal.
I do something similar in iTunes. I have a playlist set up to include all songs in the library which haven't been played within the last "x" days (where I adjust "x" periodically to manage the size of the list.) Then I run random shuffle on that playlist (which, of course, drops songs as soon as they're played.) This counteracts both the effects of normal distribution (some songs will get played more than others, if the full list is available for full-time shuffle) and my own tendency to find something I like and then play it until I hate it.
The drawback is that I get unnaturally involved in the results of this random sorting - how long it's taking, for example, to get all the "last played in 2003" songs off the list. (Four still to go.)
Posted by: pjm | April 14, 2004 at 11:25 AM