Photo by Dereck Von. Album art by Mati Klarwein
The roll-out of Spotify 0.4.3 over the last two or three weeks has been received with a whole lot of understandable hooplah, so it seems only right to add to the noise. Well, actually, not quite; there are plenty of reviews of the service out there, so I'm not going to write yet another one, but rather respond to Spotify's mobile service, a pre-0.4.3 iPhone/Android application which has been available for some time now to paying subscribers… but which, to my shame, I've only just got round to using properly (yet which is, for reasons I hope will become clear, far more revolutionary than the recent service enhancements).
A thought experiment revived: AKA Darbari Java on the Central Line
About a year back, Justin and I were approached to put together a session for some conference or the other. I put forward a presentation called something like "AKA Darbari Java on the Central line." For one reason or another the session never happened.
The idea for it had come from a tiny epiphany some time over the previous winter. Since 2000, I've been pretty much convinced that the download "revolution" was little more than a stop gap on the way to the bright future which was the "celestial jukebox". The shelves in my apartment groaning the under combined weight of LPs, CDs and cassettes notwithstanding, it seemed pretty obvious to me that the future of music consumption (music being, as an accident of smallish file size, the coalmine canary of media) was about access rather than ownership. Indeed, to this day I don't really know what it means to "own" a digital file. Yes, yes, I know what it means legally or in business terms (broadly, at any rate); but not what it really "means". Comedian Peter Serafinowicz, as I write, is the guest editor of Gizmodo; and in a hugely insightful piece, Why I Steal Movies… even the Ones I'm In, he remembered Frank Zappa saying that communism would fail because people like to own stuff. I would add a little nuance here. I think people like to own things. LPs, books, DVDs… these are all things. Digital files are mere stuff and I'm unconvinced many people really want to own stuff at all. More about Zappa anon.
All of which said, pretty much a decade on I still had some reservations about my own conviction that streaming and access were to be the future; all of these reservations were technical and practical rather than, shall we say, philosophical, but for all that, they were very real. Some time around January 09 I'm on London Underground's Central Line, way, way out of mobile phone range and listening to AKA Darbari Java/Magic Realism on my iPhone. ADJ is a wonderful 1983 album by the hugely influential cult composer and trumpeter Jon Hassell; it's a huge personal favourite, and its complex, musically ambiguous evocation of hinted-at tropical interzones is a perfect anecdote to the dehumanising bum rush that is tube travel in London.
So a thought occurred to me; for the celestial jukebox to be a viable reality, I'd need to be able to call up AKA Darbari Java while on the tube - and stream it there and them. It struck me as a pretty good test case, as it would have to assume some fundamental things were in place, namely:
Catalogue availability: Despite the album's brilliance and influence it's hardly a well-known release - nor a great seller. I know this from bitter personal experience; while a catalogue manager at EMI Virgin during the 90s I had a regular fight on my hands to keep it and the rest of Hassell's EG albums off the deletions list for which it qualified on the basis of its poor sales. Yet the great jukebox in the cloud would have to "stock" it.
Rights: Just to be clear on this previous point, I'm talking about a legal streaming service here, so for such a service to work assumes quite some IPR framework being up and running.
Ubiquitous broadband connectivity: Remember, I'm several hundred feet underground at this point. While some of the world's metro systems - Paris and Tokyo among them - have cellular coverage on them, I wonder just how long it'll be be London's massive and ancient tube network will be so equipped.
Metadata: A jukebox - oh sod it, let's just call it a server - compendious enough to include catalogue as deep as ADJ would sure as hell need a robust metadata structure in place.
Ever not been able to find something in your own iTunes collection? Ever resorted to Cover Flow for the way it taps into your visual memory when you're struggling to find that one track? Or got your music collection scattered across several PCs and outboard hard-drives? That's your collection. Let's assume that the juke-box-in-sky is a single service provided by one company. In reality I suspect there will be multiple suppliers and that the real game will be in smart aggregators which give the user the veneer of a singular service, but that's a subject for another time. Let's assume that the service(s) holds 80% of the world's recorded content although I expect that it'll be much lower for a very long time (Long Tail advocates - of whom I'm broadly, if increasingly unfashionably one - tend to misunderestimate just how much music has been recorded globally in the past hundred years). Now imagine trying to find a specific track in that lot; it will only be possible if a combination of a rigorous taxonomy, data and metadata standards and (as a possible safety net) a social-tagging schema are in place. And I stress rigorous in there.
Spotify Mobile: does it stack up, then?
When the little thought experiment outlined above occurred to me I confess that while I did believe that such services were an inevitability, I didn't think even a partially equivalent one was round the corner. But Spotify Mobile was, and I recently tried the "experiment" out.
I live in Brighton, on England's south coast. It's about an hour from London by train, and a combination of hilly terrain, a lot of tunnels and (I suspect) poorly rolled out infrastructure mean that cellular connectivity is very poor most of the journey. Especially, it seems, if you're on O2, as the swearing and muttering of Brighton's iPhone-wielding commuters will attest. So it's not the Central Line, but for the purposes of this, it's a good stand in.
As far as I can tell, the Spotify mobile streaming service maintains the illusion of a continuous service through a smart combination of buffering and caching. So I arrived at the station a tad early and dialled up AKA Darbari Java to make sure enough of it had been shoved down the pipe (I already knew from home-based listening that it was available).
I kicked the album off as we set off and the experience was a pretty satisfying one. I had, I think, three serious drop-outs during the album's playback; it's hardly like playing it off your own drive but it's impressive - to me at any rate - that this is do-able in any way at all. So how does the service stack up given the criteria laid out above?
Catalogue availability: Well, I was able to get the album on which I'd pinned the whole thought experiment, so in that regard, this was a total success. But I have to say, a year or more into my Spotify subscription, I'm still puzzled about a lot of the gaps in their catalogue. Some of the gaps are plainly down to some combination of artist, label and management refusenikery. The usual suspects are absent: Zeppelin, say, and the Beatles.
(A personal bugbear is the complete lack of Zappa. I assume this is down to the family refusing to clear the catalogue. If so, it a disastrous move. Zappa is precisely the kind of artist whose reputation could be enhanced by the new music communications ecosystem; the idea that a musically inquisitive teenager could dial up One Size Fits All or Lumpy Gravy - and have the attendant inspiration time bombs set off - is a marvellous one. It speaks to the heart of some of the copyright term-related issues raised by the recent third centenary celebration of the Queen Anne Statute that a family's concerns about the (not unreasonable) posthumous benefit from an artist's sales might get in the way of that artist's continued contribution to the tradition. (Those very term-related issues were somewhat disingenuously avoided by James Murdoch in his recent speech to UCL’s Centre for Digital Humanities on the matter, of course.) As it stands, I fear that Zappa is going to pass into musical history as an eccentric footnote, rather than up there with Miles Davis and Stravinsky as an utterly original creator of a unique vision. But I digress.)
No, the real puzzlers for me are the inconsistencies in Spotify's catalogue: I can see the possible reasons why Hassell's Fascinoma isn't in the Spotify catalogue; presumably Water Lily Acoustics haven't cleared it. But why are there, say, about a couple of dozen Keith Jarrett ECM albums from a pool of, I dunno, 60 plus. The sheer unpredictability and uneven-ness makes the paid-for service still a no-go for some of my friends.
Rights: There's little I can add on this; there's been much speculation and comment on how Spotify is clearing material and the paying for its use. It's plain from my analysis above that rights remain a minefield for the organisation. But as for this little experiment, well, again, ADJ had been cleared so it passes muster on that front. We have to assume Hassell is somehow seeing income from my plays.
Ubiquitous broadband connectivity: Well, the way Spotify mobile works intrinsically accepts that for the moment, that's simply not a reality - and isn't looking like being anytime soon. Instead, the service neatly sidesteps the whole issue. As I said above, the solution doesn't provide perfect results, but it's impressive for all that.
And lastly…
Metadata: Again, I found ADJ easily and quickly, so for the purposes of this test: flying colours. But, but… I have noticed a lot of weirdness in Spotify's metadata. I mentioned the ECM label above. As for many listeners of my generation, I suspect, ECM was the first "jazz" label I followed and for the last year I've regularly pulled up favourite albums from my teens which I have "only" on vinyl. On many occasions I've noticed that albums are simply misattributed. A case in point: last time I looked, Silent Feet by Eberhard Weber's Colours - my very first jazz album - was listed as by John Marshall, the group's drummer. I don't bring this up as an aggrieved jazz nerd (no, really, I don't) but rather because I almost gave up on finding the album at all, having scanned through the Weber inventory. It's precisely this kind of misattribution - essentially a metadata FAIL, as the kids say - which can seriously undermine the celestial jukebox functionality. And problems of findability are only going to multiply as Spotify's catalogue grows and grows.
The final analysis…
… is frankly that the service "passed" the test, in that I was able to stream Aka Darbari Java/Magic Realism, almost without break, if not on the Central Line then at least on the Brighton-London commute - like I said, almost as poor in terms of mobile reception. But in a sense, the other thoughts the whole experiment prompted for me were more salient. Spotify remains some way from being the great jukebox in the sky - for reasons which are in many ways down to conditions beyond its control, notably catalogue clearance and mobile infrastructure. But it's a good deal closer than anyone had any right to expect at this point, I think. In a sense, whether Spotify succeeds or fails at this stage is irrelevant (unless you're one of its employees or shareholders, of course). Its real importance is in showing us the potential for future of mobile music services.
A couple of footnotes
Although I wanted to take a look here specifically at Spotify Mobile, it would seem strange not to acknowledge and briefly comment on developments around Spotify 0.4.3.
User collection Integration: This is without doubt great. It's not a game-changer like the mobile service, but it's fantastic for all that. Since the upgrade, I confess that I haven't once used the iTunes console to listen to music at home, and building playlists which can now get round holes in either my own collection or the Spotify inventory has been a lot of fun.
Sharing: I understand the business sense of taking the Facebook route with networking, but I truly wish they hadn't. I've recently committed Facebook suicide - about which I'll write at some point - so I effectively have no one with whom to share playlists. Can we please do everything in our power to avoid a world in which FB is the world's default community? I'll just continue to send out links of my fabulous playlists as that works fine at the moment.
Recommendations: Still not a Spotify strength, I'm afraid Last remains, for my money, market leaders on that front. Further Last and Spotify integration as per the Spotify scrobbling has gotta be where its at!
Charging models: There are a bunch of different ways to pay for the service now - or not. PaidContent have outlined them clearly, while the BBC's Rory Cellan-Jones smartly asks what they really tell us about the performance of the company. I'll refrain from commenting myself other than to say I am continually surprised at just how many of my friends and colleagues don't think the service is worth a tenner a month. What will it take?!
Wednesday, April 18, 2012 at 6:28PM 