Universal helping make music universal

I am already pretty excited to see Universal Music Group throwing its considerable weight behind making the web more open and useful - it's even more gratifying to note that this work is being spearheaded by former members of the team I was privileged enough to lead at the BBC in a similar initiative. Let me catch my breath and explain.

The project that more than any other defined my former role as Interactive Editor for Music at the BBC was that of the BBC artist pages (here's an example)- creating, with clickable tracklists (another example), a central glue for all of the BBC's branded music offerings (radio stations, programmes, festivals etc) that served at once as navigation and aggregation (and, we hoped, a good SEO strategy to boot). What's more, we were keen from the start to make these pages last, and so we adopted the web-scale identifiers of MusicBrainz to ensure their longevity, as well as their interoperability with other parts of the Internet. I have bored on about all this in several places.

Musicbrainz furnished a ready set of metadata about artists and their discographies that also enabled us to scrape Wikipedia for biographies and pull in relevant BBC News stories (by means of inferring Musicbrainz IDs via links to official artist sites). It also theoretically made us interoperable with a number of digital music services - notably last.fm. But one of the frustrations of the project was the fact that the data in Musicbrainz, being based on commercially available releases, was only as good as the ability of its community of contributors to get hold of and input that data.

Like Wikipedia's editors, the MusicBrainz community are numerous, passionate, smart, quick, and largely self-regulating. But the thing about being a broadcaster is that we routinely got our hands on new releases - and indeed releases by commercially "new" artists - before they were sold, and therefore before the hive mind of MusicBrainz could get a purchase on them. The solution seemed obvious to me - get the labels to input their data directly into MusicBrainz (either directly or via APIs), meaning that they had control over the data that's out there on the Internet, as well as a ready-made structure and repository for their own data (plus it would save an awful lot of data entry by the BBC). But that in turn yielded a problem of its own: how to get the labels to play? I confess it's not a problem I was able to solve during my tenure.

It seems the solution was staring me in the face all along if I had only had eyes to see: embed a cadre of very clever people in the labels and let them do the heavy lifting. Step forward James Cowdery and Martyn Davies, both of whom I was lucky enough to have in my team at the BBC. James, as Universal Music UK's Innovation Manager, has been working with his predecessor Martyn, now evil genius behind Six Two Productions, to develop the Artist Gateway, using MusicBrainz IDs to create pages populated with data from Universal as well as from elsewhere on the web.

Now I should acknolwedge here that there are no current plans to include pre-release tracklist data in the Artist Gateway. Personally I think that's a shame, but I understand that established models are slow to change. But this is a huge step forward nonetheless towards creating a common language for music metadata online. 

A more general caveat is also appropriate: clearly I am bringing a rather warped and personal interpretation to this project, which is of course driven by the strategic needs of Universal rather than my preferences and whims. Plus it's early days (the Artist Gateway isn't launched yet), and I can't claim detailed inside knowledge of its likely contents or technical architecture. But the very fact that it's happening is heartening indeed.

Using linked data to build context around news

In a post a few weeks ago on the excellent R4isStatic, the BBC's Paul Rissen outlined an inspiring vision of how news could be enhanced by embracing the possibilities of the web as a medium. Hold on, I hear you cry, surely news has been there, done that and is now hurrying off behind a paywall again? Hasn't the web transformed the way organisations distribute news, people consume it and even, through social media, the way that news is gathered in the first place? Well, yes, yes and yes - but.

Paul makes the point that while the inputs and the outputs might have changed, the product itself remains remarkably similar. A story appears, without context, designed to simplify and dramatise a set of issues enough that readers, viewers or listeners will be quickly satisfied. Paul outlines how a data-driven, semantic approach to news might situate it in a wider and more meaningful context, drawing out the latent power of the web to deepen and enrich our understanding. (In so doing he has some trenchant observations to make about how it's the conservatism of journalists rather than new technology that's responsible for our supposedly shrinking attention spans.) I urge you to read the whole post and subscribe to the blog.

And I mention it now because yesterday the Guardian announced that its Open Platform Content API will allow users to query its data by MusicBrainz IDs and ISBN numbers. (I have to declare an interest; having overseen the BBC's adoption of MusicBrainz IDs as Interactive Editor for Music, I'm delighted to see another major content player take them up. Like any other standards, open data standards like MusicBrainz stand or fall by their spread and adoption.)

It's worth pointing out that this is essentially a developer-facing initiative, as a glance at the FAQs will make clear. So what's the connection with Paul Rissen's vision of a linked-data-driven news? One objection to that vision is that there is already plenty of context provided by the manual linking and proprietary tagging carried out by journalists now. And the Guardian's link-up with MusicBrainz and ISBN numbers doesn't immediately change the content or links on their own pages, which usually provide a great deal of context, again manually, in particular in the music area.

But the context that exists now rarely transcends the format of a point-to-point link or index page, and is very largely contained within originating organisations, limiting its scope really to change or broaden readers' understanding of concepts involved.

What the Guardian has done will start to enable third parties to build automatically on the context provided by their stories and the metadata around them; and (in theory at least) vice versa, pointing the way towards a much more integrated web of information around topics that might fundamentally change the experience and expectations of news consumers. And (again in theory at least), context could move beyond point-to-point links and begin to model visualisations of the information around concepts like, say, artists or books. (We did this very thing at the BBC - albeit in a very rudimentary way - around the concept of the play count for artists).

Of course this all raises a whole set of questions about business models. Paywalls represent a fundamental challenge to the viability of building open data around news (and vice versa). And thoughts about revenue models around linked open data are in their infancy. So it's a good job our own Simon is in a position where he can move some of these conversations on.