Newsletter #98 May: How To Solve A Problem Like Music Discovery?
14 May 2013
At the end of last year, when asked what was the most important issue that the streaming market had to address in 2013, Spotify’s Daniel Ek replied: “The abundance of choice. How do you make sense out of 20 million songs?”
Of course he has a point – how can anyone possibly make sense of 20 million songs? Assuming an average song time of 4:15 minutes, that’s around 1.4 million hours or 160 years of music, roughly 2.5 times the average human lifespan.
Where we once, in a time of scarcity, pored over new releases and record sleeves, giving them our undivided attention, these days the human brain is simply incapable of processing the tsunami of new music (amongst all the other media ‘content’ out there). Throw in the endless oceans of user-generated content on SoundCloud and YouTube and we, the end user, can feel like we are drowning.
For the technology industry, this situation represents ‘a consumer problem that must be solved’. Ten years ago it was the Music Genome Project (a programme that later morphed in the hugely successful Pandora) and Last.FM. Today, it is a stated aim of Spotify and various apps built around streaming – Apple (with iTunes’ Genius, which surely would be applied to iRadio) and social media giants like Twitter – to second-guess your tastes and serve you music you didn’t know you wanted to hear. There are nuances to all of these services and offerings of course, but the message is that many users aren’t capable of navigation and curating large music catalogues by themselves. There is indeed a problem, but its size and nature, and some of the proposed solutions, especially those where human agency is supplanted by algorithms, can be called into question.
Is people’s inability to make sense of 20 million songs really such a bad thing? There used to be, and arguably still is, a certain romanticism about people discovering culture they weren’t previously aware of in a piecemeal and disorganised way. Tech watchers use the word ‘serendipity’ to describe those happy accidents that lead to people tripping over new music that they then come to adore. This author fondly remembers misspelling something on a certain music download program back in 1999, resulting in a lifelong love of Chilly Gonzales, for instance. Theoretically, recommendation algorithms are supposed to increase these occurrences but in a controlled environment, with the important side effect of keeping the user glued to the platform/program itself. But is the joy of serendipity lost in a controlled system?
That said, every week, thousands of tracks are added to services like Spotify, albums of every genre – in terms of presentation and navigation, digital catalogues flatten the entire history of music into a homogenised mush, with only boring old text to distinguish one track from another. Various Spotify apps are attempting to counteract this with visual, interactive experiences that offer interesting historical and cultural context in many cases.
Another question is: are there really that many people who want their listening choices decided by a computer? On the evidence of Pandora, the answer is: yes. Will the solutions help tech-savvy music fans? Maybe, maybe not. It is difficult to believe there are that many of them who currently feel underserved by music recommendations via social media, tastemaker mixes, end of year lists, curated playlists, charts, blogs etc.
In any event, major moves are occurring in the music recommendation area. Earlier this month, Spotify announced the purchase of its own recommendation service with the acquisition of Tunigo, one of the app developers that blossomed as part of Spotify’s drive to evolve into a platform/ecosystem similar to Facebook.
On April 18th, Twitter’s company blog boldly announced that the company would be ‘changing the way people find music’, with the launch of its #music service. This news is the outcome of Twitter’s purchase of the critically-acclaimed Australian service We Are Hunted, whose website (closed following acquisition) and Spotify app (still running) was an elegant way for people to stay on top of cool new music online. The resulting #music app certainly got off to an inauspicious start, although it is sure to be improved upon. After a test drive was offered to a handful of excitable US celebrities, when real people finally got to use it, they were mostly non-plussed. You are either recommended music you already knew (Like Kanye? You’ll love Jay-Z) or by people you followed for something distinctly other than their music tastes. Bonnie Cha’s review for All Things D highlighted this issue: “I only follow people if they have interesting things to say…I don’t want to be forced to follow people just to get music recommendations.”
It is unclear who is the target of the drive for the technology-enabled music discovery solution. Is it about serving less-technologically inclined people, as Pandora appears to do very successfully? Or is it just addressing a spurious ‘problem’ for music-fanatic, tech-enabled early adopters? In other words, geeks catering to geeks who are already very capable at finding a million ways to throw up a million music suggestions.
One suspects that real musos, technologically able or not, continue to get their recommendations from the tried and tested sources – blogs, magazines, DJs, record stores, tastemakers, friends and peers. Where technology makes this easier – Pitchfork, Songza and Spotify’s social features and apps are good online examples – that is all well and good. For everyone else, there’s something called the radio.
Editorial by Tom Quillfeldt