Donut Age: America's Donut Magazine

Mix-o-matic redux

I first wrote about the Mix-o-matic, the iTunes smart playlist that serves up the vast majority of my music listening, back in 2007. It's still my go-to playlist, but it has evolved quite a bit in the past five years. It's hard to imagine a geekier topic than yet another exploration of the intricacies of my playlist system, but because the Mix-o-matic plays such a central role in my listening habits, it's hard for me to write about music without referencing it, and if I am going to do that, I figure I should at least be referencing the current incarnation of it and not its ancient ancestor.

The original Mix-o-matic basically just combined the most recent additions to my library with a selection of Good Songs I Have Neglected (rated 3+ stars and not played recently). As detailed in that old post, a limitation of the Mix-o-matic was that it ignored unrated songs (of which I had many), and for a while I experimented with variant (Mix-o-matic Plus) that added a third source to the mix—Listen More, a selection of unrated songs with a low play count—the idea being to get those songs into the rotation so that I could actually form an opinion on them. There was also a Rate These Already! playlist that collected unrated songs with higher play counts, and I would occasionally force myself to listen that playlist and assign ratings.

This arrangement was alright for listening purposes, but I wasn't really making much progress in reducing the number of unrated songs in my library. There were still a couple thousand, or about 20% of my total library, in late 2007. (Yes, I have—sporadically!—kept records on this sort of thing. Shut up.) Eventually, I concluded that one of my core assumptions—that I shouldn't rate a song until I felt like I had a reasonably firm opinion about it—was at the heart of the problem. That approach worked well enough on the extreme ends (songs I clearly loved or hated), but it led to a lot of hesitation in the middle range. Specifically, I was reluctant to give anything a two stars because that would permanently exclude it from the all-important Mix-o-matic rotation. Since there was no practical difference between a two- a one-star rating, I largely ignored the former as a category, opting instead to be a little over-generous with my threes or to just leave songs unrated forever (hence the need for the Listen More/Rate These Already playlists). The system was inelegant, and that irked me.

(At this point, it may come as no surprise if confess that the part of teaching I found almost unbearable was grading, and for much the same reason. It is easy enough to recognize excellent and poor work and to assign grades accordingly, but the whole middle range is rather fuzzily defined and mostly seen as punitive to varying degrees. But that's another story.)

My breakthrough came somewhere around the middle of 2008 when it occurred to me that I could give two stars meaning by using it as my ‘default’ rating. I had previously thought of a two-star rating as being a definitive statement that something was ‘not good’. With this new approach, it would be more of a neutral rating: not good, not bad, just there. If all songs began as twos, any other rating would represent a conscious decision. Downgrading a song to one star would remain, as it always had been, an active statement of ‘I don't like this, and I don't want it popping up unexpected when I'm on shuffle’ (one-star songs still have some redeeming value, though; if they were truly terrible I would just delete them entirely). Conversely, an upgrade to to three (or more) stars would be an affirmation: ‘This is good, let's make sure I hear it again.’ And if a song didn't particularly move me one way or the other, it would just stay a two, which seemed fair.

Using two stars in this way appealed to my sense of elegance. I could sum up each rating in a one or two words:

  • ★: (Barely) Tolerable
  • ★★: Indifferent
  • ★★★: Good
  • ★★★★: Very Good
  • ★★★★★: Perfect

I must admit, though, the idea of suddenly moving a couple thousand songs from Unrated to Two Star status kind of terrified me. I recall carefully backing up my whole library—not just Music folder but also the big XML file that holds all the playlists and such—just before the Big Change. I was half-convinced I would decide it was a horrible mistake and would have to try and revert my whole library to its pre-change state.

As it turns out, I've never second-guessed that decision, but the change did require some further rejiggering of my playlists. The Mix-o-matic list still only kept three-star and higher songs in rotation, so all these new (and any future) to-star songs were were still effectively in limbo. I didn't want to put them on equal footing with higher-rated songs. These weren't ‘good’ songs after all, but neither were they necessarily ‘bad’ songs, so I didn't want them to be completely excluded like one-star songs. The solution was, naturally, yet another smart play list: Reconsider:

The 'Reconsider' smart playlist criteria

This playlist looks at all the two-star songs, excludes a few oddball genres, and then grabs the fifty I've listened to least recently. (The Skips criterion is in there, in theory, to weed out songs that I have really decided are thoroughly mediocre, but not actually bad enough to warrant a one-star rating. In practice, though, I almost never skip songs—only a few songs in my library have even three skips—so this filter really has no effect.) The master Mix-o-matic list then shuffles the Reconsider group in alongside the New Arrivals (everything added in the last, usually, six weeks) and 100 assorted Good Songs (the playlist is still called Good Songs I Have Neglected, but somewhere along the way—probably because my library was getting so big—I changed its selection criterion from least recently played to simply random). And that's it, the new and improved Mix-o-matic.

The relative proportions obviously depend on the size of the New Arrivals group, but usually it works out to about equal parts ‘new stuff’ and ‘old stuff’, with the latter part always being two-thirds "good songs" and one-third songs being given another chance to make an impression on me (sometimes they do and get bumped up to the Good Songs pool or consigned to one-star oblivion; often they don't and get sent back to the end of the Reconsider queue). If I've gone on some buying spree (like when I run across a used CD shop) or through an especially dry spell, the proportions can get thrown out of whack, and I will sometimes make temporary changes to one or more of the feeder lists. By and large, though, the Mix-o-matic continues to serve me well as an all-purpose source of musical variety.