"Any sequence of sounds perceived as pleasing or harmonious" is one dictionary's definition of "music" - which immediately raises the question: perceived by whom? If beauty is in the eye of the beholder, beautiful music is in the ear of the listener, and as long as music has been around, one man's meat has been another man's "turn down that bloody racket".
And that's before you factor in music as fashion statement, expression of deepest feelings and identity badge. There's no need to wear the artfully distressed denim of an indie fan or the sneer'n'safety pins of a punk rocker to tell the world who you are, because a stated preference for anyone from James Blunt to Kraftwerk says as much about you as your haircut, shoes and car combined.
In fact, the music we affiliate ourselves to works as an instant likability gauge - strangers have been bonding over bands for years. But if music is so closely linked to our personalities, how can something as artificial as a website possibly determine our taste? Are we so predictable that a database can look past our quirky human foibles and identify, catalogue and shelve us into handy preconceived people-shaped compartments, according to our likes and dislikes? Surely not. Pop is a diverse beast, and it's not unusual to find The White Stripes and Kylie sharing a shelf.
Yet pinpointing our predilections is exactly what a flurry of new internet sites purport to do. Billed as "music discovery services", these music recommendation sites are geared towards widening our musical horizons. Sites such as Musicmobs.com, Pandora.com and Gnoosic.com reckon that, with a bare minimum of information, they will know exactly where you're coming from and, more important, promise to point you in the right direction along your musical journey. In the words of Tony Padilla, the founder of the two-year-old Musicmobs, they aim to "revolutionise the way people find music".
That's some ambition. But can they actually live up to it? Both Musicmobs and Gnoosic invite you to type in three artist names, from which they form their recommendations. As someone in the privileged position of receiving free CDs through the post every day (there's no better way to discover new music than that), I'd hope I'm more likely than most to know whether I actually already do like the sites' recommendations.
To give the websites a fighting chance, I opt to type in Arcade Fire, Elbow and Sigur Rós - three bands hitting out of similar, winsome rock ballparks and responsible for three of this year's best albums. Not surprisingly, Gnoosic draws up a fairly predictable list of 11 bands who are heavy on emotional post-rock. But, while I can see why it thinks I might appreciate Mogwai and Godspeed You Black Emperor, the fact is that - having seen both live and heard their albums - neither are among my favourites.
I put the same three bands into Musicmobs and, within moments, it offers another 47 acts it reckons I might want to check out; a percentage rating denotes the degree of musical similarity. Disturbingly, this time the list is a pretty accurate snapshot of what rocks my musical boat. It even suggests that I listen to the wonderful but little-known Scots Camera Obscura, an underrated old fave of mine, among other much-loved bands such as British Sea Power, Kings of Leon and Stars in its musically broad recommendation. Suddenly I realise I'm not quite as unique as I had thought.
Padilla and Gnoosic's founder Marek Gibney are reluctant to reveal the secrets of their sites' machinery, but they admit that the recommendations are based on information gathered from their site users. Musicmobs members publish their personal playlists, which are then put through "complicated collaborative filtering techniques," Padilla explains, to yield the recommendations. Gnoosic uses what Gibney calls "a unique approach to artificial intelligence" to assimilate its user feedback.
Techie jargon aside, what both sites do, in essence, is to determine taste by collating and analysing what other people like. Which is exactly what the granddaddy of recommendation sites, the online retailer Amazon, has been doing for years. The only difference is that with Amazon, those responsible for providing the "customers who bought this also bought" feedback for everything they sell, have actually put their money where their mouth is. But, while entering my bands does encourage me towards some bands I do like (and unexpectedly points me towards the great new Rough Trade signing, The Bell Orchestra), the results of these two sites seem obvious, impersonal and not specially accurate, compared to the range offered by others.
With Pandora.com (probably the most sophisticated, but not yet fully available in the UK), you do feel it's a tailor-made service. Set up by Tim Westergren in 2000, the site employs a small army of musicologists to analyse a song according to 400 criteria. "It's intense work," says Westergren, but it's necessary to fulfil the objective: "Seeking to understand the precise combination of elements that make each song unique so we can make it easier for people to find new music to enjoy based on music they already like."
Type in a band or track, and Pandora compiles a radio playlist of like-minded tunes based on specifics such as tonal or rhythmic similarities, vocal qualities and harmonic structure. It's clever, but its suggestions at times seem to stick too rigidly to the brief - not so much broadening your horizons as just deepening them.
In the end, the clinical results of a computer program will never match the special buzz of actually making your own musical discovery by hearing something on the radio, online or at a gig. Visiting these websites is like employing an intuitive personal shopper at a massive music department store - or having an cool older brother who reckons he knows what you like. It's a bit of a cheat. But the websites are fun - and they might just turn you on to something new.
Tested by Joel Pott, of Athlete
What music are you into? Architecture in Helsinki, at the moment, and Death Cab For Cutie and Arcade Fire; beautiful, glitchy electronica and singer-songwriters like Sufjan Stevens and Willy Mason.
How do you discover new music? Lots of record shopping and impulse buying.
Pott wants bands like: Sufjan Stevens, Arcade Fire, Death Cab For Cutie.
Gnoosic suggests: Xiu Xiu, The Books, Iron and Wine, Devendra Banhart, Architecture in Helsinki.
Pott says: Architecture in Helsinki? There you go! I'll have to check out Xiu Xiu. I like The Books, but I've not made my mind up about Devendra Banhart.
Verdict: This could be a good find. I'd definitely check out these bands. It seems we're predictable beasts.
Tested by Sister Bliss, of Faithless
What music are you into? Everything except metal. I'm interested in electronic bands. I like beats, but I'm just as happy with an intimate acoustic album.
How do you discover new music? Mainly through friends, plus I know some great little record shops.
Sister Bliss wants bands like: Jamie Liddell, The Editors, Goldfrapp.
Musicmobs suggests: Röyksopp, Gus Gus, Múm, Calexico, Boards of Canada, Four Tet, Le Tigre.
Sister Bliss says: I really love all those bands. I wanted Gus Gus to tour with us and I discovered Múm through a friend just last week. I've heard of Le Tigre but not heard their music. I must check them out.
Verdict: They've nailed my musical taste - it is pretty accurate. How very clever.
Tested by Andy Burrows, of Razorlight
What music are you into? Quite conventional; simple chord and melody stuff, good songs. I've always been into The Beatles, and Elton John from 1969 to 1975.
How do you discover new music? The old ways; music magazines, radio, gigs.
Burrows wants to hear: songs like "Karma Chameleon" by Culture Club.
Pandora suggests: "Come Clean" by Culture Club; "Bouncing Off the Walls" by Matthew Wilder, "Skin Deep" by Cher and more.
Burrows says: I like "Come Clean", but the Matthew Wilder track is dodgy. It's picked up on that 1980s genre, but not the sound I like. And Cher - blimey!
Verdict: I can see what Pandora's done. The site's cool, but I'd rather discover new music for myself.
Tested by Nina Persson, of The Cardigans
What music are you into? I don't stick to one genre. Most of the music I listen to is melancholic. The only thing I don't listen to is jazz. Neil Young and The Pretenders are real influences.
How do you discover new music? From friends.
Persson wants bands like: A Girl Called Eddie.
Amazon suggests: Feist and Rosie Brown.
Persson says: Feist has been recommended to me. I've heard some songs and she sounds amazing. Rosie Brown? Not sure; I'm not easily finagled into new stuff. If a friend I trusted recommended it, I might buy it.
Verdict? Not terribly helpful. I'm really lucky to have a bunch of friends whose opinions and taste I trust. But the recommendations point you in the right direction. I've used them when buying books.Reuse content