Feed Sign in with OpenID OpenID

Simon Willison’s Weblog

Missed opportunity

Apple are missing out on a huge opportunity. I’ve bought 198 songs through the iTunes music store now, and iTunes has access to the other music that I’ve imported from my own CD collection. I want to discover new music—why doesn’t iTunes look at what I listen to, match it against buying habits tracked through the store and give me an Amazon style “people who like the music you listen to also liked...”. Privacy concerns could be avoided by having the recommendation feature off by default—I’d turn it on in a heart-beat.

Posted 8th April 2004 at 12:09 am

This is Missed opportunity by Simon Willison, posted on 8th April 2004.

View blog reactions

Next: A myriad of markup systems

Previous: Glastonbury screw-up

11 comments

  1. They are probably already working on it. While I'm sure privacy issues are there (think limewire downloads), it wouldn't be a big deal to get around them. Probably the main reason this doesn't already exist is a core functionality argument.

    Apple is very big on the whole "let's make sure we're doing the core business right, then we'll hit the other features". While apple could probably just license this technology for the store from someone like amazon, they will undoubtedly want to do something impressive with it that no one has thought of. And while the iTunes music store seems essentially static as far as features go, they are probably working on a large number of features for upcoming releases, of which this is probably relatively minor.

    It is a great idea though. But then again, it took 3 years to get finder labels back into OS X.

    Joe Mullins - 8th April 2004 00:36 - #

  2. Oh hell yeah. I'd love that feature. Good idea and I hope they are working on that as we speak.

    Keith - 8th April 2004 00:46 - #

  3. You might be interested in this... Cheers!

    Michele - 8th April 2004 00:50 - #

  4. "You might be interested in this...Cheers!"

    I was just thinking the samething. I saw it at iPodHacks.

    Grant - 8th April 2004 02:35 - #

  5. Or maybe this?

    Manuzhai - 8th April 2004 05:15 - #

  6. I totally agree with that comment. And the real kicker is that it wouldn't be that hard to implement from a coding standpoint either. I mean, they already keep track of what you have purchased and more likely than not what you have previewed (and if not, that's no more than a couple lines of code). So, it would be a breeze to show you songs/artists that are similar when you login.

    select disctinct song from songtable where songid in ((select songid from purchasedsongs where songid in ((select songid from purchasedsongs where userid = [your user id] )) ))

    Or something similar. I don't have a db with those tables to play this against, but the general gist is there. Just use that to populate a "Other artists you might be interested in" section and it's all good.

    Dave Giffin - 8th April 2004 05:30 - #

  7. For my shoutcast station (started pretty much as a lan-coffee induced hallucination ;p) I spent, oh maybe 10 minutes, putting up something similar for requested songs. An example info: System of a Down - The Metro.mp3. A little crossreferencing song titles against requesters and it kinda sorta works. Usefullness & reliability would prob be better with a larger sample size though.

    mike - 8th April 2004 08:10 - #

  8. Never mind that missed opportunity; we aren't even able to buy tunes in the UK.

    Ronan - 8th April 2004 13:17 - #

  9. You might also be interested in this. Lets you upload your iTunes library and discover ppl with similar tastes. (I just discovered it a couple of days ago, funny you post about that topic now)

    Gerrit Kaiser - 8th April 2004 22:24 - #

  10. Yes, Musicmobs is quite similar to what you are talking about. Maybe Apple should work with them.

    Tetsuo - 9th April 2004 00:36 - #

  11. The Rio Karma mp3 player keeps track of how many times you listen to each song and when you last listened to them. If the iPod had something like that, Apple could give ratings not just as "Other people with this song bought X" (which has little information), but as "Other people with this song bought X and really like it, but many didn't like Y."

    sfb - 9th April 2004 17:30 - #

Comments are closed.

Previously hosted at http://simon.incutio.com/archive/2004/04/08/opportunity

A django site