The End of Scrobbling – A Farewell to Last.fm

Digital music has been a fascination of mine since the turn of the millennium.  Audioscrobbler came into being in 2002 while I was in college, and the thought of sharing my listening with a global network of musical peers was exhilarating.

Audioscrobbler merged with Last.fm in 2005, taking the social element of music to a whole new level.  There were forums to discuss listening trends, metadata analysis and recommendation engines… all while independent blogging exploded onto the scene in a flood of obscure music fetishism.

In the years since I admittedly lost touch with the service and dropped off the scrobbling radar to focus on personal relationships, collecting unscrobbleable LPs, and developing my career.  As the summer of 2015 came to a close my life was settling up nicely – I left Windows for Linux, I have a fiance, a fantastic career, and I’ve just purchased my first home.

With these stations of life secure, my mind returned to the world of scrobbling and the possibilities of merging big data and my own hyper-specific musical tastes.  I developed a ~500 day plan to scrobble every track from my library 24 hours a day for over a year to submit every title toward Last.fm’s recommendation engine.  Surely a library of over 110,000 tracks would produce some intriguing results!

But this evening, I logged into Last.fm and looked around to find that the site has retired all of its original functions.  The social forums are closed.  The “neighborhood” of your peers is now inaccessible.  The homepage offers only a most-popular-globally-this-week roster plastered with “Uptown Funk” and other predictable tracks.

The Wikipedia spelled out what I’d missed – CBS had acquired Last.fm for £140 million in 2009.  Wasting no time, in February of that year the service handed listener data over to the RIAA over concerns about a then-unreleased U2 album.  By 2010 the service closed the custom radio feature, (again over licensing issues) and in early 2015 they partnered with Spotify, further crippling the usability of the site.

But the nail in the coffin came in August of this year with their fully-overhauled website.  It received almost universally negative criticism from its users, who cited broken and missing features.

Given the new light of this information, I’m terminating the full-library scrobble project and saying farewell to Last.fm.  Still, I shall not mourn the loss for long.  The social function of digital music has experienced a parallel evolution in the world of private forums and closed groups on social media sites like Facebook.

A magnificent record I discovered thanks to a Facebook Record Community

Every morning I’m greeted with “now-spinning” rare vinyl treasures and independent music reviews which top anything you’d find from a recommendation engine.  One user from South Korea offered nearly 40 daily installments of records from his Tangerine Dream collection, each accompanied by a custom write-up on the featured release.

Private tracker communities, classic bulletin board systems, and other social structures of the web continue to serve as a brilliant resource for musical discovery.  Last.fm served us well during a pivotal time in the age of digital media, and it will be missed, but we’ll carry on.

RIP Last.fm
2002 – 2015

The Last.fm Project

My server is down for maintenance for the next 16 hours.  It was a perfect opportunity to begin my next long term music project.

When Innerspace Labs first switched to the cloud, I used the web-based RacksandTags service through my OrangeCD DB to create an index of all track information from my library.  Collections on the service can be searched by artist, album, or track, but lacks support for 2nd level organization like genre clustering, playlists, and other more valuable data points.

I later switched to Discogs.com.  Discogs offers real time market value assessment of your collection, but only supports physical media.  I was also disappointed to find that user-generated category foldersare not presently shareable with other users.  
As I prepared for the downtime last night, I realized that I hadn’t given Last.fm a shot since I wiped my account clean in 2014.  That year I scrobbled 30,000 tracks, but was frustrated that there was no way to submit all my library’s data without playing every track in real time.

My goal was to explore the service’s recommendation engine, and my library data would likely produce some valuable results.

So last night, I went to work.  I quickly realized that the best approach would be to queue all 100,000+ tracks and to scrobble them in order of ascending track duration.  I organized the songs into four pools of nearly equal size.  Below is a map of my library based upon these four classes – less than five minutes, less than ten minutes, less than thirty minutes, and up to 24 hours.

As the largest batch was that of the shortest tracks, there would be the greatest (and fastest) return from scrobbling these first.

I charted the play duration of each of these groupings to see what sort of timetable I’d be looking at for project completion.

Graphing the duration of each grouping clearly demonstrates that this was in fact the best course of action.

I began scrobbling immediately for the first time in a year.  Once the project is complete I’ll share some of the resulting recommendation data Last.fm provides.  I’m looking forward to it!

Happy Labor Day weekend everyone!