Last.fm Seasonal Impact Indices

Everyone’s expe­rien­ced that thing where you’re lis­te­ning to something, and you think to your­self, “Holy shit does this remind me of fall 2004.” How strongly cer­tain music is corre­la­ted with cer­tain periods of your life depends on many things, inc­lu­ding but pro­bably not limi­ted to when you first heard it, when you first liked it, and when your lis­te­ning to it was most highly con­cen­tra­ted. So, for ins­tance, in my case, most Des­tro­yer albums will recall times and pla­ces that are vague at best, and that depend mostly upon first expo­sure rather than con­cen­tra­tion — this as a result of the fact that I lis­ten to every Des­tro­yer album all the time, approximately.

Blueboy’s Uni­sex, on the other hand, will pro­bably always remind me of the win­ter of 2006–7, as I lis­te­ned to it for the first time that sea­son, nine addi­tio­nal times within that sea­son (rac­king up about 150 tracks lis­te­ned, accor­ding to Last.fm), and vir­tually never again once spring hit.

Ever since I began sub­mit­ting lis­te­ning data to Last.fm in Novem­ber of 2004, I’ve won­de­red whether I’d ever enjoy direct access to all those num­bers. Then came Last.fm Extra Stats, mer­ci­fully collec­ting all my lis­te­ning data for me in a tab-separated file that can be pulled into Excel and mani­pu­la­ted to my heart’s con­tent. Here, as a small exam­ple of the data, are my top ten artists (by tracks lis­te­ned) from win­ter 2006–7, along with total lis­tens for each artist (since Novem­ber 2004) (now that I’m finally get­ting around to publishing this post, all the follo­wing data is very old):

Win­ter 2006–7
Artist	Win­ter (S) ↓	Total (T)
Trans Am	163	163
Blue­boy	148	163
The Lucks­miths	69	105
Rata­tat	50	126
The Moldy Peaches	49	51
White Flight	36	41
Tele­vi­sion Personalities	35	35
Beach House	35	64
Revol­ving Paint Dream	32	58
RJD2	31	52

Now for some metho­do­logy. We only have two num­bers avai­la­ble to us here; let’s call them S (sea­so­nal total) and T (total total). As I said, the psycho­lo­gi­cal impact of a band/record over a given time period depends on both the amount of lis­te­ning during that period (S) and the per­cen­tage of total lis­te­ning within that period (S/T). Neither figure is good enough on its own; I may have lis­te­ned to Des­tro­yer 102 times this sum­mer (I did), but that’s insig­ni­fi­cant given my 1,500 ove­rall lis­tens. And I may have only lis­te­ned to The Lady­bug Tran­sis­tor this sum­mer (yiel­ding a 100% con­cen­tra­tion), but 3 lis­tens is hardly enough to make an impact.

But, since both figu­res are posi­tive indi­ca­tors of impact, we can just mul­tiply them (S²/T). This would give Lady­bug Tran­sis­tor a sum­mer 2007 impact index of 3, and Des­tro­yer, 6.92.

Let’s see how this mul­ti­pli­ca­tion affects winter’s top ten artists by tracks played:

Win­ter 2006–7
Artist	Win­ter (S) ↓	Total (T)	Impact (S2/T)
Trans Am	163	163	163.00
Blue­boy	148	163	134.38
The Lucks­miths	69	105	45.34
Rata­tat	50	126	19.84
The Moldy Peaches	49	51	47.08
White Flight	36	41	31.61
Tele­vi­sion Personalities	35	35	35.00
Beach House	35	64	19.14
Revol­ving Paint Dream	32	58	17.66
RJD2	31	52	18.48

Most notably, Rata­tat is knoc­ked down a few notches (on account of how con­sis­tently I lis­ten to them), The Moldy Peaches go up a few spots — and if we reor­der the entire list (of 78 artists) by impact index, a cou­ple more eke into the top ten:

Win­ter 10 Impacters
Artist	Impact (S2/T) ↓
Trans Am	163.00
Blue­boy	134.38
The Moldy Peaches	47.08
The Lucks­miths	45.34
Tele­vi­sion Personalities	35.00
White Flight	31.61
The Six Parts Seven	24.00
Rata­tat	19.84
Loney, Dear	19.70
Andrew Bird	19.31

This last table is what we’re really after, theo­re­ti­cally giving us the ten artists who will most strongly remind me of the win­ter of 2006–7. It seems a bit boring at this point, since it’s pretty close in its results to the stan­dard top-ten-by-listens chart; but the metho­do­logy is sound, and under more extreme lis­te­ning con­di­tions (e.g., a Des­tro­yer obses­sion), it will prove to be a relia­ble indi­ca­tor of impact.

And don’t for­get, as time pas­ses and I lis­ten to these artists more, their season-specific impacts will drop. For ins­tance, as things tur­ned out, I lis­te­ned to Tele­vi­sion Per­so­na­li­ties hea­vily over the spring, inc­rea­sing their T value, and redu­cing their impact in the above table (which uses old data). As a result of my TVP spring lis­te­ning, their win­ter index drop­ped dra­ma­ti­cally from 35 to 7.95 — their spring index, as of Octo­ber 5, 2007, is 91.95.

Depic­ting this analy­sis visually could go seve­ral ways; somehow I want to see a pie chart. Making one from every artist would be imprac­ti­cal, lea­ving a circle with a hun­dred or more sli­vers. Taking the top ten artists feels insen­si­tive to the quirks of any given sea­son; what if I don’t even lis­ten to ten artists that sea­son, for ins­tance? What if I lis­ten to a thou­sand? I could select only those artists who have indi­ces grea­ter than, say, 20; but this also feels arbi­trary and vul­ne­ra­ble to sea­so­nal quirks.

I think a pretty safe threshold above which to choose artists is at 50% of the sum of all impact values. In other words: the sum of the impact indi­ces of winter’s artists is 783. If I take artists off the top of the list until I have half that, I stop after the fourth artist, The Moldy Peaches. Admit­tedly, a four-sector pie chart is kinda dull, but I need to be con­sis­tent in the crea­tion of these charts so that impacts are accu­ra­tely repre­sen­ted from sea­son to season.

So, roughly…here are what the last three sea­sons of 2007 “meant” to me:

What I like about these is that they are not only inter­nally con­sis­tent, but con­sis­tent with each other; Ariel Pink will remind me of the spring of 2007 more than M.I.A. will remind me of the sum­mer, as repre­sen­ted by a lar­ger sec­tor in his season’s pie. I could simi­larly gene­rate these with a focus on albums, which would be slightly more spe­ci­fic and alter the results dra­ma­ti­cally. I know, for ins­tance, that all my sum­mer 2007 lis­te­ning to M.I.A. was off her new record, songs that neces­sa­rily didn’t con­tri­bute to any prior M.I.A. lis­te­ning — so, Kala, were it to have a sec­tor in an album pie, would have one much lar­ger than the less spe­ci­fic “M.I.A.” one.

Gran­ted, there are more than a few pro­blems with this whole endea­vor, and here are the major three:

1. Last.fm does not track things that I lis­ten to on CD/vinyl. This is a serious pro­blem, as the things I purchase tend to be the things I like the most, and go totally unac­coun­ted for.
2. The num­ber of tracks you’ve lis­te­ned to by an artist is only kind of rela­ted to how much time you’ve spent lis­te­ning to them. 4 tracks can fill 2 CDs (Lift Your Skinny Fists), or you can cram 91 tracks into 40 minu­tes of music (Irre­vo­cably Over­dri­ven).
3. iPod lis­tens are not pro­perly docu­men­ted. Even if I remem­ber to report them to Last.fm with Audio­Pod, mul­ti­ple lis­tens on a sin­gle track only get coun­ted as one — this is par­ti­cu­larly dama­ging on long trips, for instance.

The (S²/T) for­mula may have further appli­ca­tions, as I’ll touch on in a follow-up to my ori­gi­nal post about alpha­be­ti­za­tion and digi­tal music library navi­ga­tion.