Several times before, I have made reference to unusually-appropriate or unexpectedly-amusing juxtapositions of songs on the work box, which contains an iTunes install of some 5,650 tracks. The mathematics of the process would seem pretty simple, as Tam notes:
So I’m driving down the freeway yesterday, listening to Moby’s cover of “New Dawn Fades” on the iPod, best known as the tune from the Heat soundtrack playing when Hanna pulls over McAuley, and the next song the thing offers up in shuffle mode is the original Joy Division recording of “New Dawn Fades”. What are the odds of that?
After reflection, she determined that it was 1/512, which implies 512 tracks on the iPod. In comments, TJIC noted that since this can happen in two ways cover followed by original, or original followed by cover we’re looking at more like 2/512, or 1/256.
Which makes perfect sense if the shuffle is perfectly random. Given the fact that computers generally don’t do a bang-up job on randomness in the first place, I figure there’s some chink in its armor.
A 2007 experiment conducted under the auspices of Cnet seemed to suggest that commercial considerations were a factor: “Artists and singles purchased through iTunes were played more frequently than those that were not.” I haven’t noticed any such tendency, though my own shuffle contains fewer than 150 tracks actually purchased from the iTunes Store. (There are about as many tracks bought from Amazon’s MP3 storefront, which the Amazon installer automagically moves into iTunes after purchase; I haven’t noticed any favoritism one way or another with these items.)
What I’m wondering about, though, is if the shuffle has enough smarts to select the beginning of song B because it goes well with the ending of song A. In my mix-tape days, I put out well over 300 cassettes, and in my opinion, the two best transitions I ever did were:
- Stevie Wonder’s “Sir Duke” to Badfinger’s “Day After Day”;
- Petula Clark’s “Kiss Me Goodbye” to Jimi Hendrix’ “Purple Haze”.
The shuffle has indeed coughed up one of these. “Sir Duke” was playing, and I observed, “Ideally, the next track to come up would be ‘Day After Day’.”
Which began, right on cue. I am going to rip “Kiss Me Goodbye” today and add it to the playlist, to see what happens.
My own preferences will tend to complicate the mathematics: I run a floating playlist called “Randomator” (!) which pulls 500 nonclassical tracks based on, depending on the mood of the moment, either least-recent airing or fewest plays. The list remains at 500; one song is played, the next one takes its place. For least-recent airing, this is pretty simple, but for fewest plays, it’s a little more complex, because no song has more than 23 plays at the moment, meaning lots of ties. And the shuffle breaks ties, apparently, in reverse order by artist’s name: the “5” Royales and the 5th Dimension, then the 4 Seasons, more numbers through 10cc or 10 Years, and then ZZ Top, Warren Zevon, and so on, all the way up to Abba. (Before you ask: Aaron Neville is sorted with the N’s.) And this confounds me: I might be in some particular mood and will find the song that just came up incompatible with that mood, so I’ll press the Next button. The shuffle, often as not, will bring it back within six hours.
And I still haven’t figured out how that Genius gizmo works, though I admit to being impressed with the machinations that produce the Genius Mixes. I freely concede everything that people hate about iTunes: it’s slow to load, it’s cumbersome to operate, and it chews up memory faster than stoners wolf down Cheetos. But having gotten a handle on most of its features, I wouldn’t be without it at work, anyway. At home, I seldom have the urge to run nine or ten hours of continuous music.