This post concerns an experimental internal-BBC-only project designed to allow users to bookmark, tag and rate songs they hear on the radio using their mobile phone. It was developed by Matt Webb and myself (with Gavin Bell, Graham Beale and Jason Cowlam) earlier this year. Although the project is a BBC project, all the speculation and theorising around the edges is my own and does not necessarily represent the opinion of my department or the BBC in general.
We have more television stations than we have time to watch, more radio programmes than we can fit in analogue frequencies, more music and film availablethan any human could consume in their lifetimes and a huge ever-growing world of information growing every day on the internet. And this is just the beginning. The next push is the archive – decades of programming coming online, lost films recovered, libraries being digitised. But the scale of even this content is dwarfed by the third push into the world of the amateurised content-creator, where potentially billions of people are putting information and media out into the world as a matter of course.
The most substantial challenge to technology creators, media creators and distributors is – then – to find ways of making this enormous and every-growing repository navigable and sensible to real people. There are substantial rewards to be found in finding ways to help people find their way around this space – and people familiar with the challenges of the web over the last ten years are in exactly the right place to work out what these navigable mechanisms are likely to look like. But you don’t only have to create the navigation to reap the rewards – the organisations that can supply the right metadata, supplementary and structured relationships about and around their media will be the ones that will survive most easily inside this new ecology.
There’s also one more major challenge. Current media distributors and large-scale media creators are going to find themselves suddenly operating in a market of peer creators, where hundreds of people can create and interact and respond to the media around them. The network is already a challenge to broadcast – people who use the internet a lot use television less – but this is a new challenge. It’s a challenge of participation – where one-to-many broadcast-style content has to figure out how to find new ways of getting their ‘audience’ involved. This is a challenge that’s all over the place – and it’s a problem of bandwidth. How does one show or product or team respond to and respect the input of hundreds of thousands of individuals, and reflect it in what they make? If you’re last.fm it’s easy – you give everyone something different. But if you’re a popular content creator with one outward channel that’s the same for everyone, things get a little harder. How will they adapt?
This is the world that a few of us went to ETech this year to talk about. Mr Biddulph and Paul Hammond talked about BBC Radio’s current offerings and live work, particularly digital radio, on-demand streams and RadioPlayer and the famous Ten-Hour Takeover. Meanwhile Mr Webb and I talked about some more experimental work we were doing (in collboration with people like Gavin Bell) on the assumption that navigation, interaction and user-creativity were the core media issues of the next twenty years. We talked particularly about two projects: Group Listening and Phonetags. At the time, I promised to post something about the latter project at the time, but I never got around to it. After renewed interest from the FooCamp crowd, I thought I’d do it now.
Radio networks have always been interactive, but they suffer from bottlenecks. If you ask people to vote in a poll and then report the results, then you are to an extent reflecting your audience on air. But it’s a fairly homogenised and averaged-out view of their beliefs – pushed through a fine-meshed sieve. The variety is completely lost in the aggregation.
On the other hand, if you want to get some of the spice and individuality of the individuals concerned you can pick out specific individuals from your audience and put them on air (or mention them). Unfortunately, many individuals find the prospect of being on air more than a little intimidating – and of those that don’t, still only a fraction can actually be featured on air.
Both of these approaches have worked perfectly well for many years – but we’re now at a point where we can start thinking beyond them. So the question now is – what is beyond aggregation and lottery? What new patterns of interaction can we form around and within broadcast now that we have a networked world to hybridise with it?
Phonetags is an experimental project designed to try and help us get some purchase on some of these questions. The best way to describe it is to start off with some Principles for Effective Social Software that we developed as a result of working on the project. I’m not going to pretend that they cover everything, but they’ve proven very useful for us. We believe that for a piece of Social Software to be useful:
- Every individual should derive value from their contributions
- Every contribution should provide value to their peers as well
- The site or organisation that hosts the service should be able to derive value from the aggregate of the data and should be able to expose that value back to individuals
So this is how it works. Phonetags is about bookmarking songs you hear on the radio using your mobile phone. The way you use it is very simple. If you’re listening to a radio network (initially BBC 6 Music) and you hear a song you’d like to make a note of, you pull out your mobile phone, type an ‘X’ into an SMS (remember: X marks the song) and send the text to a BBC short-code. Later when you come to the site, you type in your mobile number into the search box to see a list of all the songs that you’ve bookmarked:
As you’ve probably already noticed, bookmarking isn’t the only thing you can do with Phonetags. You’ve typed in the ‘X’ to bookmark, but you can type in other stuff too – any words you type after the X are considered tags in the same style as Flickr and del.icio.us. You can navigate your own tags and explore other people’s tags – both in aggregate and individually as you see fit:
You can probably start to see in the latter screenshot why this stuff starts getting so valuable for us, at least. Those keywords – along with their reflected popularity – are starting to provide a pretty clear articulation of what the concept of ‘rock’ means.
Alongside the bookmarking and the tags, we added a new concept called ‘magic tags’. Basically these are special tags – like magic words – that perform some action upon the song that you’re bookmarking. At one level you could view them as nothing but compensation for a lack of UI widgets in an SMS interface, but there could be value in having tags that were both semantically interesting and also performed an action of some kind.
The tag we used in this circumstance was a simple ‘rating’ tag. If you wrote a tag of the form *one, *two, *three, *four or *five, you would mark the song as having been rated one-five stars for quality. This seemed to make a lot of sense in the music space, as it’s something people are familiar with from applications like iTunes, and you could imagine a range of circumstances where people might wish to express their opinions on songs played.
This view results in my favourite view in the entire system – that of the top-rated songs for any given ‘tag’:
A page like this exists for every tag in the system – there are pages of the top rated indie songs, pop songs, guitar songs, summer songs. You can imagine a whole range of possibilities for extending these pages to make them permanent or to atrophy with time / create weekly charts. It’s a huge mine of interesting musical information and a great way to discover new songs.
Anyway, the point of Phonetags was to try and find a different way for a user or audience member to participate in programme-making and to collaborate with one another without any of their contributions being lost, and with the value accreting over time.
In this model, a user gets value from their very first contribution – by having a song bookmarked that they can return to later. They gain extra value by being able to keep track of and comment upon the songs that they’re listening to – and when they do so, everyone else starts gaining value as well.
The peer benefit is in music discovery and navigation. There’s an incredible amount of new music being produced all the time. Our increased access to it means – in principle – that we should be able to find music that we felt more appropriately suited us, but the sheer volume makes it hard to explore. With a service like phonetags, an individual can start exploring music by axes of quality, or by keywords or by discovering people with similar tastes to themselves. And it gets updated in pretty much real-time.
Radio DJs gain a little bit of this experience too, in that they’re able now to operate as a peer in this exchange – tracking a bit more rapidly how well people are responding to songs, and using the live site as a way of mining for songs on any given theme (give me ‘happy’ songs, songs about ‘summer’, songs about ‘mum’, songs about ‘fruit’). They can also court reactions from their audience – rate all the songs in this week’s shows and we’ll play the best at the end of the week…
But it’s behind the scenes that I think the most substantial value could be created. We’re getting in incredible metadata on music that we simply didn’t have before – metadata and descriptive (emotive!) keywords that we can analyse and chop up and use as the basis for all kinds of other navigational systems. This is metadata that is often sorely lacking and could help us enormously in the future.
Anyway, I’d be delighted to hear any comments or thoughts that anyone has on Phonetags. All the images above can be clicked on if you want to see a larger version. If you want to contact me, then it’s tom {at} the name of this website (as usual). At the moment, we’re testing this particular version of the service inside the BBC (it’s available to all BBC staff to use so if you want the URL, then just let me know). The project is unlikely to be released to the public in its current form – but we’re using it as a way of testing out some of these concepts and approaches – some of which will probably manifest in upcoming products in one way or another.
And just to give you the disclaimer one more time: Phonetags was developed by Tom Coates and Matt Webb with Gavin Bell, Jason Cowlam and Graham Beale. However, all the opinions expressed in this piece should be considered as my own personal take on the developing media landscape, and not necessarily those of my employers or the department in which I work.
35 replies on “Reinventing Radio: On Phonetags…”
What do i get again? 🙂 Sorry to be obtuse, but as a user what’s the benefit to me?
I know it’s a bit weird to make a feature request for a piece of software that I haven’t used and couldn’t yet if I wanted to, but: I’d like to be able to use this via email – texts cost money and I often listen to 6 on my computer.
Looks great, by the way. 🙂
Stewart – you get the ability to bookmark songs you hear on the radio (say if you don’t know what a song is, you can bookmark it and come and look it up later), and you get to organise your bookmarks and send in your feedback about songs you’re listening to as well. Plus you get the benefit of everyone else’s tags and ratings to explore.
Interesting idea – hope you bring it to States. I’m the founder of a qtags, which follows the same concept – bookmarking advertising so that you can remember it and learn more later. http://www.qtags.com or blog: http://www.allaboutqtags.typepad.com
This is really, really cool stuff. I just hope you can do an Apple-esque job of explaining how the listener can make use of this service without overloading them with features. I think radio would make this particuarly tricky to make clear, butI definately think there’s potential if it does get off the ground.
P.S. Love that you put Jeff Buckely – Everybody… in one of those screengrabs.
Well,
one of the problem I see with SMS interfaces is that the user has to write his message in some sort of a “script language” – e. g. in this case “X *one Rock Cool Favourite”
For a radio station, I believe, a big part in making this success is to communicate to the listeners *how* they can use the service. An option might be sending them to a website, where they can look it up, but that’s not really mobile, is it.
So what I would opt for is probably that users can request a free MMS or SMS with instructions. (MMS would also allow them a visual preview of the site with screenshots, to communicate their benefit). Also, a WAP interface would be nice. Then I wouldn’t have to remember the “sript language”, but I could just select from a dropdown the rating and then fill in tags in text fields.
Best, Bjoern
Why not also let them phone a voice line for a second or two, and optionally speak a number from 1 to 5? Presumably you can still tell what the user’s number is, and they don’t have to laboriously type out the message.
What occurs to me is a time delay factor.
I might be listening to a song for a minute or two before I decide I want to tag it. By the time I’ve pulled my phone out of my pocket, got into the messaging, composed a new message (after thinking a bit about emotive keywords I will relate to later) I send the message, but it’s too late, Celine Dion has started playing, and now not only have I failed to tag that excellent Foo Fighters song I just heard, I have added the keywords “Rock” “Visceral” and “Grind” to My Heart Will Go On, and I look like a complete tool to anyone who stumbles across my tagging.
So, how do you overcome this issue, being that songs on the radio aren’t normally more than about 3 minutes, it’s a fairly small window in which to solicit the input from the user.
Is there the option to be able to review your tag online and say “no, I meant the song that played BEFORE that one” – and what if you really aren’t sure, would it be possible to expand the service to include streaming samples of the songs? Then it has some real value to me. I can remind myself of the songs I tagged three months ago when I can’t for the life of me remember which one it was that went “I’m the crazy frog!” and I want to go and buy the CD without having to endure the embarassing task of singing the song as I remember it to an amused yet annoyed music store employee.
Interesting idea, but the cost of texting is a barrier that would stop me using it. One thing that could be a problem, though, is the possibility of tag-bombing by record companies, artists or other agencies which would put an unwanted and unrepresentative emphasis on some records. Or, worsem the sort of thing we’ve all seen with wikis where spam companies screw the whole thing with irrelevant tags. Mind you, this hasn’t happened with Flickr, has it, or is it that it’s not popped on to their radar yet?
Hi Tom,
It sounds a little like the Shazam system, I wonder if it were more commercial (i.e. not the BBC) if you could download the tune from the website, or even to a mobile direct.
I think it could also work for podcasts, tagging a news item for instance, or a football report, or just a phone in to listen to again later, and finding other similar reports or programmes. Of course the message would be slightly different, channel and then a time stamp..
I think it’s a cool idea by the way!
Stew
With regards to the time-delay factor, yes – Matt Webb spotted that one and built in a really cool partial delay which means that for the first twenty or so seconds of a song any SMS received bookmarks the previous one. Another feature which we didn’t build in this prototype but which we accepted would probably be useful was a way of editing and correcting your bookmarks on the website – particularly with a ‘wrong song? previous / next’ interface of some kind.
To Card-shark – spamming is always a problem, I agree, although there are substantial anti-spam benefits in using a mobile phone as an interface and requiring people to only tag when the song is playing. The latter means that you have to employ people to lurk around a radio waiting patiently, and the former means that you have a relatively solid way of uniquely identifying people and stopping people sending in multiple entries. If they act up you can simply block the phone number, and it would be extremely difficult for any individual organisation to bring together enough phones to compete with potentially tens of thousands of real-world users.
The fact of the radio playlist also helps in this regard. The songs are tagged – not the time they were broadcast, so if it is well-played over a period of time it will have many opportunities to be tagged by individuals, which should minimise the spamming effect from people trying to game the system.
Exciting stuff. There seems to be something missing, though – with del.icio.us, you get to click the links you find by browsing tags, on Flickr, you get to see the photos, but on the PhoneTags site, there doesn’t seem to be a way to listen to the songs you find. A good second best would be an indication of, say, the show(s) that have played a given song in the last week, but that doesn’t seem to be there either. It’s like the system is set up to be incredibly useful, but with the one key element that users are interested in left out.
Sorry to snipe, it really is a fascinating project, but I can imagine being frustrated by having to fire up the iTMS or a filesharing application once the PhoneTags system had guided me to a new song or artist.
Jack – well of course there’s no reason why we couldn’t have a way for people to listen to the songs played on the site – at least for a good percentage of them we already post links to short samples on our profile pages: Pixies profile. And then there’s all the stuff that Mark Thompson talked about over the weekend about the desire of the BBC to help people complete their music journey with a purchase off our site. There are all kinds of possibilities for the BBC to start setting itself up as an impartial way of navigating around music and helping people discover music they might like – supported by several national music radio stations… But all of that’s a fair way off…
This sounds fantastic. Although I wouldn’t use the tagging stuff from my mobile (although would happily add them online later), the ability to bookmark a song is so so useful. The amount of times I hear a song on 6Music on my walk to work, and I miss what it was, or forget what it was, or forget to look it up when I get to work.
I would kill puppies to use this service.
Well done team, bonuses all round.
I like it, though I’m not in the UK so I can’t use it.
I agree that it would be nice to be able to click on a song and hear or even buy it. I don’t know how quick the BBC would be to partner up as an iTunes affiliate, but that seems like a good match. Radio’s good at driving singles, and the iTMS is good at selling them. Seems logical that I’d be able to bookmark a song I liked and then go buy it when I get home.
Very interesting. I have two thoughts on this. First, I already tag and rate all my music in iTunes, and it would be nice to be able to run a script and upload all of my tags and ratings to the site to gain the aggregated information, e.g. similar music I might like. Second, I’d like to see the recommendations move beyond similar music to something that incorporates time into the mix. Music tastes tend to shift over time, and it would be interesting and useful if the system could not only identify other users with similar tastes right now, but also identify other users who used to have similar tastes, and have now moved on to new music tastes. And the opposite would be useful for discovering musical roots. The first example that comes to mind is something like: “You like the Beatles – Check out Paul McCartney, the new Beatles” and “You like Paul McCartney – Check out the Beatles, the original Paul McCartney” and “You like the Beatles – Check out Chuck Berry, the original Beatles.” And so on. This wouldn’t be people who are listening to both bands now, but rather people who once listened to one band, and have now moved on to the other.
With music listening and/or purchase in place, this could get very interesting indeed, I think – I see from Mark Thompson’s speech at the EITF that the BBC are already talking to Universal with regard to opening up archive material which they have rights to.
In fact, looking at this in the context of that speech, with its emphasis on openness, universal access, interactivity and partnering with rights holders, plus what you say about the BBC providing ‘an impartial way of navigating around music and helping people discover music’, it seems like the BBC is shifting (even further) towards becoming a new kind of organisation, way beyond being a broadcasting corporation.
I wonder what the implications are for things like the license fee, the BBC’s role as a public service broadcaster, and the charter renewal? This sort of merging of broadcasting, user-led content and straightforward commercial activities puts it in a very odd position. The sort of position that, say, existing music download services might have a bit of a problem with.
Sorry, drifted a bit far off the topic of PhoneTags there.
I remember on John Peel shows I always used to have a pen and paper handy to write down all the bands I wanted to lookup later.
Some points: You need a preview, even if its 10 or 20 second, for shows like onemusic you might have to take a recording of the show and chop it up into little bits. I also would like to see what’s playing now (across the network) and be able to tag stuff via the site.
If your explaining this to man on the street their reaction might be “What’s the point?” it will be interesting to see how you do explain it in a simple way and further more developing radio features around it.
I can see how shows such as Zane/Scott Mills would benefit, as you say rather than having a “text vote” users could tag songs they like throughout the show and the one with the most could be played as a “U Turn”.
By happy coincidence, there was someone in the office just last week talking about phone tags although I’m not sure whether they were talking about PhoneTags, if you see what I mean.
Anyway, the really interesting thing about PhoneTags is how many of the same ideas have made it into Millionsofgames.com. We needed to build a casual games tagging site purely out of neccesity (so many games to use and play, so little time) and in MOG we seem to come up with many comparable ideas. The ‘Principles for Effective Social Software’ you’ve outlined are close to what we had in mind when we started to alter MOG to make it more user friendly, especially to an audience who aren’t likely to have used del.icio.us or flickr. The same thing applies to ideas like special tags and ratings.
We’ve also looked at the overall metadata that tagging games gives us and realised that we needed to atrophy the popularity of certain lists. MOG now has two distinct measures of popularity – tags and clicks – both of which we combine to give a popularity ranking for each game and an activity ranking for each user.
http://www.millionsofgames.com/game/ology/1384/
http://www.millionsofgames.com/games/pop/
http://www.millionsofgames.com/mog/activity/
The obvious next step is to start creating lists of games to highlight their popularity over time, which also exists but we need more users before it becomes widely representative of anything but our own internal preferences.
http://www.millionsofgames.com/2005/
Back to PhoneTags though – have you any plans to use it on anything other than music? Being able to tag Mark Kemode’s film reviews would be something I could see myself using just as much, if not more.
This sounds ace…..finally something that might get my boss to let me listen to the radio at work, after all I’d be helping the BBC become even better :o) (can I have the link so I can try and persuade him)
Really interesting stuff… though the cost of sending each text might count against it. Re. the metadata: there’s something similar in operation on the allmusic guide (much used as reference material by music critics): http://www.allmusic.com/cg/amg.dll?p=amg&sql=75:
Good luck with this.
Related to music similarity discovery, you should check out Pandora from the Music Genome Project.
I have a number of comments detailed at http://www.corante.com/getreal/archives/2005/08/31/tom_coates_on_socializing_radio.php.
Conclusion: “In the final analysis, I agree that radio will need to be socialized, but I am not sure that a broadcast medium like conventional radio can make that jump. Certainly, the BBC or other broadcast media would like us to provide all those tags and ratings, but why would we hand over all that metadata to them, since it may not actually change our personal experience of the radio a bit? They can’t fracture the broadcast into a gazillion streams, with those tagged “downtempo” finding their way to me. It just provides a “tagalicious” means for the BBC to profile their market, as opposed to a way to have a many-to-many communication setup.”
I’m very fond of the basic idea behind this, but I have to wonder what the actual utility will be. Not *could* be – *will* be.
There is an enormous amount of listener feedback that can be gained through mechanisms like this, but I’m not convinced at all that that will end up increasing the quality of music being broadcast.
First of all, obviously, one can only rate music that one hears on a compatible station, which subsets your choice dramatically. Compatibility with something like Audioscrobbler (just to take a random example) would help to create a “people out there like this” database.
Secondly, SMSing your choices from your mobile is not ideal at all. It costs money and it takes time. I don’t know who is going to be sitting in front of their radios or computers with streaming audio, phone in hand, just waiting for the opportunity to do this, but it certainly isn’t anyone I know, even the real music geeks. SMS is the ideal public communication medium for low bandwidth but for one song every five or so minutes, it’s not ideal.
Thirdly, there needs to be the belief that sending data will result in an actual change in playlist. Even assuming that we are dealing here with the BBC rather than commercial radio, this belief is going to be hard to form. Over the last few years I’ve seen an increasing level of distrust in the BBC’s dedication to broadcasting good new music, engendered by, well, the BBC – and I remember when Radio One was considered a station that you could listen to for new music. Filling it with clone zoo radio song promoters doesn’t engender optimism. Why would anyone bother? While this is still going on it’s a distinct handicap. This really is the major barrier. How are you going to enforce this?
Oh dear – I seem to have slipped into a “bloody Radio One” rant there. Please ignore or remove last paragraph.
In any case, the above relates more to the entry of metadata. I think the base idea of bookmarking is very valuable (another John Peel + notepad user here) though texts are still rather expensive for regular users. Multiple access points would seem to be the way to go; a button on the BBC Radio Player popup or widget, or a seperate popup window, widget or small dedicated program would let anyone with net access use the service. Those who are listening, say, in the office, without full net access could bookmark via email – which would work for smartish phones as well and would be cheaper than SMS. Finally, if all you’ve got is a phone, you’ve still got the SMS option.
I think multiple access points would be key to the overall success of the scheme, which generally sounds seriously promising. What’s needed is a profile for each user whereby they can enter one or two phone numbers and a couple of email addresses from which all the individual’s tags are collated.
Paying for an SMS would be the last resort for many people, with the ability to email from home or work. But if you’re on the bus or whatever, desperate not to miss the chance, it’s worth 10p. Speaking as someone who endlessly repeats the artist and titles of decent new songs until I can find some paper, and more-often-than-not turns on the radio midway through the best songs and misses the info, this would be invaluable!
As a former indie radio show maintainer, I love the idea. Although I used to keep my playlist up to date and as thourough as possible, it was the feedback that I lacked, and having my audience rank what they got to hear would be very cool option.
Myself, as a listener, I’d also love to have this option, and the cost of texting wouldnt stop me – because my experience tells me that to come across a tune that I relaly like is a rare thing, and I’d hate to miss the opportunity to learn what’s being played!
Dmitri from the city of Izhevsk, Russia
Looks nice, but it just seems the exact same as last.fm. Well, with star ratings and obviously not having a direct feed from the user’s desktop as to the music they’re listening to and tagging.
Which is to say that it’s good (and obviously developed before the recent, shiny new last.fm relaunch), but like others have mentioned, the cost of entry is comparatively expensive.
what a fantabulous idea! I find however that I do almost all of my radio-listening in the car so it might pose quite a danger to others if I were to tag a song via sms. (although I can’t tell you how many times I’ve tried to find pen and paper in my bag with one hand on the wheel to jot down a lyric from a song I like so I can look it up later. I think something akin to shazam where I could call in and tag a song by speaking would be ideal (at least for folks in california where we spend way too much time in our cars 🙂
It’s a great idea. How often have I wondered how the song is called… (I’m not in UK, though.) And BBC doing it hopefully takes care of most of the obvious privacy problems with it.
Most interesting I found your hint about DJs. They can get inspirations in real-time, based on their audience. Or I might search for a song fitting to a certain subject: around a discussion on radio, background music in a movie, personal gift, musical postcard.
I also fully agree with your problem analysis in the beginning. Check out ManyOne http://www.manyone.net, which I am working for.
Tom – instead of *quality* have you considered using the tag(s) as criteria against which the rating(s) are made?
I can’t believe I haven’t found this before nor heard of it. I’ve been looking for something like this (and trying to create the idea in my head) for months. Having recently moved to Amsterdam, where RDS is much more developed than in the States (Europe in general being ahead of the US), I’ve thought of the idea many times in the car. Reason being, here I get the name of the artist and sometimes the name of the song. Often times I think, hey, that’s a great song and then can’t find a way of remembering it. An X is brilliant. The station has all the timing info so knows exactly what the song is can relay that info for your future listening pleasure. Now, lets go online and integrate that song information to an amazon or itunes link so we can take it a step farther. Then, the radio station would get a portion of the profits when you purchase/download the song or album. As a matter of fact, I think radio stations have been extremely technologically lazy…this goes for the US more than here. They need to add value to the listener. Especially with the tremendous benefits of XM and other forms of satellite radio…how long will they be able to survive without any innovation. The problem though is that too many radio stations out there play a bunch of crap and don’t help you to discover new music. One that actual does it well here is Arrow Jazz (www.arrow.nl/jazz), continually helping me to discover new music. Now, I just want a way to remember and share :). If anyone wants to discuss this further please let me know, as I think combining this with qtags represents some great possiblities. X.
there was a project about 6-8 years ago doing something like that with real radio. I think a canadian radio station offered little keychain buttons that let you bookmark or infoseek songs while they were laying on this particular channel.
you might have read about it in WIRED, shame it never cought on. I was just thinking about something like that for internet radio, as anything is digital today shouldnt be a biggy to implement.
Found this link yet dont quite understand the full service, anyways keep me updated when its out there.
i think so
Phonetags is a great tool and I like it.