This article make me think about the kind of devices that will be used to store huge amounts of music.
Specifically the kind of search that this devices will allow, for me is a real trouble to deal with 200+ songs. It would be awesome to tell the device to play 8 random Jack Johnson songs, then 6 random songs from Rolling Stones Voodoo Lounge album and finally to play "U Can't Touch This", or to make it play those 15 songs in a random order.
But what happens when you don't remember the artist, album or song name?
When you can have unlimited music, it's time to think in the next ipod :P
this is a good point. i have a hard time dealing with one of songs I download or purchase. Since they are not in an album (and are often by an artist that I barely have a couple of songs from) I tend to forget about them after a while and never listen to it for a long time, until I accidentally stumble upon it. I can imagine finding songs you like out of all the songs ever recorded would be pretty tough.
tagging and user ratings will play an important role as users are given more choices.
rather than how itunes sorts your music, the model used by netflix is a bit more useful. realistically, in a few years everyone will have easy access to library's the size of netflix, available for instantaneous streaming. there is no reason to assume that music will not adopt a similar model.
Moore's Law seems like it has a limit. Could we fit 10 terrabytes of data into a device the size of a phone? 100 terrabytes? I wonder what the maximum is.
All the data won't be stored in the phone, it'll surely be streamed from somewhere. As we saw our documents moving to online Office, we'll see in few years all these big files moving away from our hard drive.
Who has the hard drive space keeping an up-to-date divx library today? I personnally prefer to download a movie when needed, p2p has become fast enough to be convenient :)
You're right. And that has interesting implications for future mobile devices. Right now, phones connect to the internet using a connection provided by your phone company. EVDO or EDGE, for example.
But how hard would it be to create a peer-to-peer internet connection? One of your devices would be connected to WiFi, but also connected to every other device in the area. And those would be connected to more devices, and so on. It would form a big chain, possibly extending several miles.
When someone requests a song, that request is passed from one device to another until it gets back to the device that's connected to WiFi. Then the song data is streamed from the internet and passed back to the original requester.
This system seems like it would be extremely slow. But, does it really matter if it takes an hour to download 10 songs? You could have a list of songs that you'd eventually like to listen to, then check back every few hours to see which ones are ready.
Mesh networking is indeed slow, but it works great for some things. I'm not sure if music is one of the however. The nice thing about iTunes music store or a fast connection to a P2P network is instant gratification. I remember dialup days when it took 15 minutes to download a song and that was painful enough. Now if I want a song I can get it in under a minute and once you get used to this it's hard to wait several hours for the songs to potentially finish downloading.
On the plus side, chances are that one of the other devices in your mesh already has a copy of the song you want (especially if it's a popular song) and if storage isn't an issue you can have a lot of locally cached data that should speed things up.
> But how hard would it be to create a peer-to-peer internet connection? One of your devices would be connected to WiFi, but also connected to every other device in the area. And those would be connected to more devices, and so on. It would form a big chain, possibly extending several miles.
I think, only fans will pay for music and go to concerts. The others will listen to music for free. I don't know anybody in Russia who pays for a music regularly. But I know a few fans who buy legal albums.
When people pay for music, they're paying for the convenience of the delivery, not for the music itself. For the average user, it's worth a dollar to avoid the relatively clunky wilderness of Limewire and BitTorrent.
Hmm...I'm afraid I don't understand why do you considering using P2P unconvenient. Cos here in Russia it is the most comfortable way to get music. Internet music market is only a few $millions for now. But even if it will grow, I doubt that people will drop P2P. Cos it easy as open StrongDC++ in your metropolian network, type an artist name and download all songs at the price $0,0004 per mb.
Maybe I'm wrong and you have different situation in US.
Just my 2 cents.
using p2p depends on what other people are sharing. I want to download some older, lesser-known artists: Damien Jurardo, Stavesacre, and the list is pretty slim. If you want something popular and recent like Linkin Park, no problem. But if your off the beaten path, it's kinda hard to find what you want
I predict it wont be about how much media you can store locally on your device at once instant, as much as how immediately accessible new media will be.
I could have an iPod with 10 petabytes of media, but if I wanted to view a picture my friend just took or listen to a song released after my last sync, the device is worthless unless I have a good means to access this new data.
I like the flood idea. While we still have limits, a simple filter that roughly gauges potential interest in the file would help make this possible to implement today.
Add an app to your iphone that discovers other iphones. It'll probably be bluetooth. Then add a mechanism to transfer files automatically between the iPhones, probably wifi. In addition, to get past the 4 or 8 GB limit, backup unplayed or unlikely-to-be-played songs to a private music repository whenever a wifi internet connection is available.
This system can be built with today's hardware, and is essentially under the radar of music companies -- more so than bittorrent.
But this really emphasizes the need for recommendation systems -- the only thing the labels are actually good at. Have you ever tried to find something you might like on bittorrent? Almost impossible. P2P apps of yore were better. You could see someones playlists. You could discover new things when grouped near things you know you like.
But that has nothing to do with quality. Especially in niche genres (and everything is going the way of the niche), quality matters. There are many failures for each success in a style.
19 comments
[ 5.6 ms ] story [ 136 ms ] threadSpecifically the kind of search that this devices will allow, for me is a real trouble to deal with 200+ songs. It would be awesome to tell the device to play 8 random Jack Johnson songs, then 6 random songs from Rolling Stones Voodoo Lounge album and finally to play "U Can't Touch This", or to make it play those 15 songs in a random order.
But what happens when you don't remember the artist, album or song name?
When you can have unlimited music, it's time to think in the next ipod :P
rather than how itunes sorts your music, the model used by netflix is a bit more useful. realistically, in a few years everyone will have easy access to library's the size of netflix, available for instantaneous streaming. there is no reason to assume that music will not adopt a similar model.
Who has the hard drive space keeping an up-to-date divx library today? I personnally prefer to download a movie when needed, p2p has become fast enough to be convenient :)
But how hard would it be to create a peer-to-peer internet connection? One of your devices would be connected to WiFi, but also connected to every other device in the area. And those would be connected to more devices, and so on. It would form a big chain, possibly extending several miles.
When someone requests a song, that request is passed from one device to another until it gets back to the device that's connected to WiFi. Then the song data is streamed from the internet and passed back to the original requester.
This system seems like it would be extremely slow. But, does it really matter if it takes an hour to download 10 songs? You could have a list of songs that you'd eventually like to listen to, then check back every few hours to see which ones are ready.
On the plus side, chances are that one of the other devices in your mesh already has a copy of the song you want (especially if it's a popular song) and if storage isn't an issue you can have a lot of locally cached data that should speed things up.
It exists! See for example http://www.fon.com/
By the way, P2P networks don't have to be slow as long as they're well designed; they can even be faster than traditional client/server.
Maybe the popular peer-to-peer options in Russia are more mass-market than the ones in the US.
I could have an iPod with 10 petabytes of media, but if I wanted to view a picture my friend just took or listen to a song released after my last sync, the device is worthless unless I have a good means to access this new data.
Add an app to your iphone that discovers other iphones. It'll probably be bluetooth. Then add a mechanism to transfer files automatically between the iPhones, probably wifi. In addition, to get past the 4 or 8 GB limit, backup unplayed or unlikely-to-be-played songs to a private music repository whenever a wifi internet connection is available.
This system can be built with today's hardware, and is essentially under the radar of music companies -- more so than bittorrent.
But this really emphasizes the need for recommendation systems -- the only thing the labels are actually good at. Have you ever tried to find something you might like on bittorrent? Almost impossible. P2P apps of yore were better. You could see someones playlists. You could discover new things when grouped near things you know you like.
But that has nothing to do with quality. Especially in niche genres (and everything is going the way of the niche), quality matters. There are many failures for each success in a style.
Any ideas on a quality detection system? :-P
It can be paralyzing.
For the hackers out there focused on recommendation engines and algorithms, PG's essay "How Art Can Be Good" is required!
http://www.paulgraham.com/goodart.html
A good question may be: How do the ideas outlined in that essay square with your music recommendation service?
[[[fwiw]]]