Alphabetization Is Not Fit for Music Libraries

Thanks snute.

The weather affects my lis­ten­ing a lot, espe­cially if it’s a big change, like wak­ing up to a gray, wet, quiet day, or the first big snow of the year. Time of day is impor­tant too. And fluc­tu­a­tions in the NASDAQ average.

Wal­ter: Actu­ally Song­bird has a Wikipedia add-on that helps to pro­vide that sort of con­text. Just the other day I was try­ing to come up with a bet­ter search solu­tion than that, as Wikipedia isn’t really very com­pre­hen­sive in the bands it cov­ers. What if I’m cur­rently lis­ten­ing to Grouper and I want to see, in one place, all her releases, notable crit­i­cal com­men­tary, related artists, etc.? I couldn’t even think of a proper “jump­ing off” point, some site that has even links to all those things. Last.fm is a decent solu­tion, as part of its func­tion is to serve as a wiki for every band, in the­ory, but you’d have to go else­where for crit­i­cal context.

Since almost every band has a MySpace page now, that’s actu­ally a good start, despite all the things I hate about MySpace. Lesser-known bands espe­cially tend to link to reviews of their albums, blog announce­ments about new releases, and pro­vide presskit-quality art­work and bio­graph­i­cal data.

There’s actu­ally a great arti­cle about what you’re describ­ing that I meant to link to in this post:

But the con­tent expe­ri­ence on the Web is crap. Go to Aquar­ium Drunk­ard, click an MP3. If you don’t get a 404, you’ll get a Save As… dia­log or the SAME GOD DAMN QUICKTIME BAR FROM 1995. OMFG. ARE YOU KIDDING ME? THIS IS ALL WE’VE ACCOMPLISHED IN 15 YEARS ON THE WEB? It makes me insane.

Thanks for the com­ment, “Me.”

There’s one sim­ple answer to your ques­tion: If you want to lis­ten to The Who, you put your mouse in that lit­tle search box in the cor­ner and type “The Who.” If you’re using an alpha­bet­i­cal list to find artists and albums also, then that’s two tools redun­dantly serv­ing the same purpose.

I’m not talk­ing about mak­ing mixes, although that’s valu­able to some peo­ple too. I, like you, am “tied to that archaic cre­ation” as well — almost all of my lis­ten­ing is by full album. In fact I even men­tion in the arti­cle that one of the short­com­ings of the most pop­u­lar media play­ers is that they don’t “think” in terms of albums.

What I’m try­ing to address is the prob­lem of hav­ing too much music, and too diverse taste. I have 1,664 albums in my library, and I’m sure there are other peo­ple with thou­sands, if not tens of thou­sands, more. I don’t always know what I want to lis­ten to; in fact, I’d wager that most of the time, I don’t know what I want to lis­ten to. I know I want to lis­ten to some­thing, and that that some­thing is in there some­where, but I don’t know what it is.

If you asked me to list, from mem­ory, all 1,664 albums in my library, I’d prob­a­bly never make it past a cou­ple hun­dred, and that would take days. If I’m in a sit­u­a­tion where I have to fig­ure out what to lis­ten to, only the artists who are most famil­iar to me will spring to mind first, but that leaves me trapped lis­ten­ing to the same things over and over. What I’m after is a sys­tem that will know bet­ter than me what I want to lis­ten to, based on my lis­ten­ing habits, artist sim­i­lar­ity, etc., and which can help me to explore the things I have that won’t spring to mind.

A short story: yes­ter­day I was perus­ing my phys­i­cal CD col­lec­tion (gasp!) to find a spe­cific album I wanted to lis­ten to while I walked to work. While look­ing for that one album, I found another album that I real­ized was more appro­pri­ate for my mood, for the weather, for the time of day, for how long it had been since I last lis­tened to it…etc. Now, why didn’t I know that’s what I wanted to hear when I first approached my CD col­lec­tion? It’s because my mind is very flawed and very finite, and can only know so much about what my library looks like.

The Who is a very well-known band, and one that I’m sure you’ve lis­tened to a lot, so it’s easy for you to know that you want to lis­ten to them. But wouldn’t it be a bet­ter sys­tem if, when you’re brows­ing (not search­ing) your library for The Who, you’d find them adja­cent to The Rolling Stones, The Kinks, and Cream, rather than Wil­son Phillips, Wilder­ness, and Wes­ley Willis?

It’s like I said in the arti­cle: If you know what you’re look­ing for, you search. If you don’t, then an alpha­bet­ized list is not the way to find it.

And I think your laun­dry anal­ogy actu­ally works in my favor. Shirts go together because they are of the same type of cloth­ing, just as The Who and The Rolling Stones are the same type of music. I would never sort my clothes by their brand, for instance, yet this is anal­o­gous to sort­ing a music library by artist. I don’t care who made the clothes, I care that they are appro­pri­ate for the weather, that they match, that I haven’t worn them recently, etc.

Thanks again for the com­ments guys.

As far as the “ton of work to cre­ate and main­tain” goes, Adam’s right in that there’s meta­data every­where. He men­tions Pan­dora, which is kind of a top-down approach to clas­si­fy­ing music: “experts” metic­u­lously cat­e­go­riz­ing every­thing they can by musi­cal “genome.” I pre­fer Last.fm, how­ever, with its empir­i­cal, bottom-up approach to clas­si­fi­ca­tion, based on track­ing actual lis­ten­ing habits. That, and they have a long his­tory of mak­ing their meta­data free and acces­si­ble with pub­lic XML ser­vices for just about every­thing. So, in short, you wouldn’t have to do any­thing; your music player would know which artists should be grouped together automatically.

I sup­pose when it comes to a mouseless/keyboardless inter­face like the PS3 (or a portable player, come to think of it), alpha­bet­i­cal lists do make search­ing eas­ier than hav­ing to “type” artists’ names; so in those cases, I’d have to con­cede that you’re right, Me. And I sup­pose what I’m describ­ing will be use­ful only to peo­ple with large libraries like myself. But for those of us who have those large libraries, min­ing our col­lec­tions is incred­i­bly frus­trat­ing, espe­cially when you real­ize that there are so many oppor­tu­ni­ties to cre­ate brows­ing envi­ron­ments that are not only more intu­itive and func­tional, but fun as well.

Adam’s also right that these kinds of inno­va­tions should be “effort­less,” yet it seems as though nobody’s inno­vat­ing. Nobody has been think­ing beyond the alpha­bet­ized list until recently, with Apple’s “Genius” fea­ture and Microsoft’s “MixView” — and both of these seem to be efforts to sell music, not pri­mar­ily to make your library more dynamic.

All it would take is a room of user-interface experts exper­i­ment­ing with dif­fer­ent mod­els to “push” your music to you in inter­est­ing new ways, based on artist sim­i­lar­ity, based on your lis­ten­ing his­tory, based on mood, based on time of day, based on time of year — hell, based on the weather and on news feeds. It wouldn’t merely be flashy; it would, I think, pro­foundly change the way you inter­act with your music library.

Thanks Shazback. One rea­son that we’ve had to stick to alpha­bet­iz­ing books is that, in most col­lec­tions, there’s only one copy of each of them. So if you were to want to sort by “genre,” you would have to make very sub­jec­tive choices in orga­ni­za­tion. What’s great about dig­i­tal col­lec­tions is that they lend them­selves to what’s called a “flat hier­ar­chy” — some­thing I talk about with regards to music else­where here and here. Items can be dupli­cated with aliases and placed under mul­ti­ple cat­e­gories. Kind of like if you’re look­ing in the new releases sec­tion at Block­buster and some­body has taken the extra step of mak­ing a shelf of old clas­sics by the same direc­tor of a new release, or with the same lead actor. It’s those rich cross-sections that are miss­ing from our music libraries.

Ide­ally under a sys­tem like this you will find The Bea­t­les next to The Beach Boys and next to Oasis. Or, if we want to get really tricky, all three of those bands will have the same X and Y val­ues, but Oasis may have a dif­fer­ent Z value as a result of their being influ­enced by ’60s pop, and not a part of it them­selves. A 3D map­ping like this makes things a lit­tle more com­pli­cated than is prac­ti­cal, but it’s worth think­ing about.

Last.fm may be rec­om­mend­ing bands to you that you don’t like, but if the Last.fm sim­i­lar­ity data were applied to your own library, then you would only be pre­sented with things you pre­sum­ably like (since you own them). And, although Last.fm may not be per­fect, it’s prob­a­bly the most thor­ough and avail­able dataset out there — and, in my opin­ion, it’s pretty damn good.

If you pre­fer to lis­ten on a song-by-song basis, then yes, I admit that this becomes a more dif­fi­cult prob­lem to han­dle. All­Mu­sic and Pan­dora do pos­sess mood-centric meta­data at the song level, but don’t make it pub­licly avail­able. With­out their coop­er­a­tion, per­sonal tools might be the only way to attack this prob­lem, such as Moody for iTunes or (the now defunct) Mood­Logic.

On the album level we only have mood-centric meta­data from All­Mu­sic. This is sub­jec­tive, of course, and even deal­ing with full albums on a mood-based level is method­olog­i­cally ques­tion­able, since many albums are all over the place in terms of mood. So we either (a) just accept that this is a fun­da­men­tal prob­lem with describ­ing albums by mood, or (b) devise some algo­rithm to apply the moods of every song on an album to the album as a whole, which would make an album’s inclu­sion under a par­tic­u­lar mood not binary — i.e., either it is or is not “sad” — but on a gra­di­ent — i.e., it is 80% “sad.”

Per­son­ally, with my hack-ish imple­men­ta­tion of these moods on the album level in foobar2000, I find that I get very good results. And though there are albums that are emo­tion­ally all over the map, there are also many that are emo­tion­ally pretty sta­ble; think about Love­less or Slanted and Enchanted. Fur­ther­more, even if every song on an album is not per­fectly described by the moods used to describe the album, I would argue that in most cases, albums as a whole, even if they vary in mood, either do or do not com­ple­ment the mood of a sit­u­a­tion. All­Mu­sic will often describe albums that are emo­tion­ally erratic as “cere­bral” or some­thing, rather than strictly “bleak.” Check out the full list of AllMusic’s mood descrip­tors and you’ll see that it’s actu­ally an admirably nuanced system.

On the artist level, this is where mood breaks down. Although sum­ming up an album by mood is a dodgy but viable endeavor, the same can’t be said for artists. Brows­ing artists by mood would be fruit­less, so I don’t really pro­pose it. This is why a 2D genre/style map is in my opin­ion the best way to orga­nize artists. Last.fm already does this with their “Islands of Music” exper­i­ment, which, as an inter­face to a desk­top music appli­ca­tion, would be bliss.

Grand­mas­ter Flash is described by All­Mu­sic as being: Bois­ter­ous, Brash, Party/Celebratory, Con­fi­dent, Bravado, Play­ful, Vis­ceral, Free­wheel­ing, Ener­getic, Gritty, Intense, Exu­ber­ant, Omi­nous, Provoca­tive, Aggres­sive, Rous­ing, Somber, Con­fronta­tional, Cathar­tic, Dra­matic, Searching.

Louis Arm­strong is described by All­Mu­sic as being: Warm, Lively, Free­wheel­ing, Care­free, Amiable/Good-Natured, Earthy, Cheer­ful, Joy­ous, Play­ful, Bois­ter­ous, Earnest, Roman­tic, Glee­ful, Rous­ing, Ener­getic, Fun, Con­fi­dent, Whim­si­cal, Exu­ber­ant, Ele­gant, Rollicking.

Where would this put these artists in AllMusic’s more gen­eral mood cat­e­gories like “Fun/Good-Natured” and “Slick/Smooth” from their Tapes­try project? I don’t know — that’s too much data to extrap­o­late a quick guess from. But even if they were neigh­bors in mood, so what? Like I said, if you have a decent mood cat­e­go­riza­tion sys­tem, genre becomes irrel­e­vant. Hell, right next to each other in this image I have cLOUD­DEAD (exper­i­men­tal hip-hop) and Cyann & Ben (psy­che­delic folk-rock) under Bleak/Cold. And I can tell you, yes, both of those albums are pretty bleak and pretty cold, and if one would com­ple­ment my mood at a given moment, there’s a good chance the other would as well, despite the two being from vastly dif­fer­ent genres.

This con­ver­sa­tion too has strayed from another major point I was try­ing to make in the arti­cle, which is that your own per­sonal play his­tory within your music player pro­vides a very rich dataset to be exploited. I myself have spent some time devel­op­ing an algo­rithm to describe what I call “Hot­ness,” which looks at how fre­quently and how recently you’ve lis­tened to any given song to give you a snap­shot of the artists that are more or less your “cur­rent favorites.” And despite being admit­tedly rudi­men­tary, it still pro­duces vastly more inter­est­ing results than sim­ple “last played” and “num­ber of plays” sort­ing or “smart playlists” (“smart” — hah!). And I’m sure some­body smarter than myself could invent far more com­pelling ways to look at your library that are based on your play history.

There are a lot of ways this could all go, and none of them is per­fect. But I think it’s impor­tant to start think­ing about this, and for peo­ple to start at least attempt­ing to cre­ate a richer, more dynamic inter­face to our libraries. I’m not an expert pro­gram­mer by any means, and am just now try­ing to verse myself more thor­oughly in Javascript and XUL to start work­ing with Song­bird. But I was hop­ing to at least inspire con­ver­sa­tion with this post, and to poten­tially reach the right per­son who both shares my ideas about what these solu­tions might look like, and knows enough to imple­ment them somehow.