Yup, you are missing something. Check out CC Mixter's tagline -- "mixversation".
Remixing is about bouncing ideas back and forth, getting in a conversational flow, a jam approach that isn't done in real-time. It's like a mailing list or a forum. Here and there you see really cool things, the rest of the time it's just people talking.
The idea of being a struggling musician is a bogus way of thinking. The practice of music is not about hustling to become a star. It's about making music constantly, and once in a while making some money on it.
By throwing nodes at the problem. The virtue of P2P is that it scales up like nothing else. The disadvantage is that you need a lot of nodes. Given enough nodes, though, there is nothing with even remotely comparable uptime.
I welcome your pea-sized mind to this earth, Brother SnakeStu. Be at peace.
Tip: there's no complicated issue here. The not-very-complicated issue you're referring to (the default copyright on anything copyrightable) is not relevant.
The sense in which you are wrong is that a license is a different thing than copyright. Your statement that a rights holder gets rights automatically is true but has no effect on the licensing situation, which is the comment you were trying to refute....relevance of demo scene to OMR, well, uh, SHIT! MY MIND JUST BLEW!...where's that eyeball... over there on the floor, hm, there it is... *pop* (back in). I dunno, can't see how that issue came into the conversation.
To make this kind of thing work, you need a lot of candidate songs to find one good song, for whatever definition of good you want. You take a massive number of songs, throw them at the wall, and a few stick. The numbers that OMR had, in the hundreds, weren't even close to the kind of numbers you needed. Webjay has about 50K songs now, and people toss in a few hundred new ones a day.
That's the way it is with collaborative filtering -- you either have a huge amount of input data or the system doesn't work at all.
(About the licensing thing, uh, yeah, well, thanks, but first of all you're wrong and second of all even if you're weren't wrong it wouldn't be relevant. The issue is that OMR couldn't use music from the demo scene because most of that music is deliberately released without any documentation describing the licensing conditions.)
The difference is in the scale of things. No flame intended, but Webjay knows about orders of magnitude more songs than OMR.
Most people in the music industry don't know anything about open licenses. They've barely heard of Linux. Their first reaction when they do hear about open licenses is to recoil.
Then, there's a huge amount of public domain music from the demo scene. But the demo scene culture doesn't use licenses at all, usually -- people just post their music without a license.
Then, there's a lot of music from areas where the GPL has no mindshare, and there's a lot of historical music from before there existed things like software licenses.
The original poster got the point of the thing exactly wrong. This isn't about unauthorized music, it's about sticking to music which is both free and authorized.
Filesharing networks are full of hit songs which are unauthorized. Out on the public/stable web there's a huge amount of unknown stuff which is authorized. The trick with Webjay and other sources mentioned in the article is finding the few great songs in the whole sea of crud. If you can do that, you can have good music which the rights holder doesn't mind you having, though you usually have to give up on name brand musicians.
There's also some kind of Gnu-ish angle that I've never been able to articulate well... It's something like -- if you *choose* to listen to music that isn't from insane powermad label types, you get (more) liberty, and if you choose to listen to non-free music, you give up liberty. That's not quite right because almost none of this music is under a free or open license, though.
FWIW I'm the author of Webjay.org and was quoted in the story.
It's not affiliated with Real in any way. Real playlists do have a bunch of functionality that Winamp and WMP playlists don't, because Real is the only major vendor with half decent (buggy but still usable) SMIL support. As a result playlist authors who depend on SMIL features also tend to have a preference for Real URLs....the alternative is proprietary MS video, which is no better.
There's a fair amount more video there, but I agree that it's too hard to find. Keep in mind that this video playlisting thing is user-driven and took me by surprise.
About your other comments on e.g. "it's just a list of mp3s", that's exactly wrong. The client side remix of multiple overlapping media objects is where the fun stuff happens. That depends on SMIL, which depends on Real, so you haven't seen what people are talking about.
(Not exactly true that you have to have Real for SMIL. You could use QuickTime, except that QT's SMIL support is painfully buggy and crash-prone. You could also use proprietary IE extensions.)
The bad buffering is because clients all suck so bad. Within the community of hackers working on this stuff nobody has taken on the really crucial job of making a client that doesn't suck. (VLC obviously doesn't suck, and all the components you'd need are there, but there's no SMIL support.)
On a technical level, what's original is that the remixing all happens on the client side. It's a *client side remix*, which is a new thing.
Check out this playlist for a fancier set of techniques, including clipping, multiple audio and video sources at the same time, and a good playlist in general. When you watch that the thing to realize is that the soundtrack is coming from one place, the picture from another, the video from another, and all of that is getting mashed together on *your* machine.
Webjay doesn't alter the original. All it does is pass along a URL with some metadata about where to start and stop. That's about as much of a modification as the height attribute on an img tag.
If pine uses lynx to view html mail, isn't that a violation? Seems like you have to pay off Eolas if you want to do things the Unix way, in lots of small modules that do one thing well.
I just gave Coleman twenty five bucks. You should too.
Contribute to Coleman online: http://colemanforsenate.com/
If you contribute, be sure to call the campaign office and say why, e.g. that this is about his stance on the RIAA. The number is (651) 556-1846. They'll take your comment very seriously when there's money behind it.
If you're a Democrat, like me, just lay down your guard and be a single issue voter for a minute. If getting slashdotted means money for politicians, they'll start taking us seriously.
Speaking as the author of the oreillynet blog, the news is that an anonymous poster revealed the vulnerability that I had given up on finding. Except that followup gossip about the exploit posted on the blog is that it's not the same one I was trying to find, meaning that there are two, but only one has been publically exposed.
You don't get it. What Relatable (among others) can do is provide a pretty good guess. What it can't do is provide facts. Guesses are useful in aggregate, as statistics, but useless in court.
Audio fingerprinting is not something like a hash function that leads to a deterministic identifier. It is more like a web search engine that finds the best fuzzy match.
If you use audio fingerprint scores in the aggregate, for example to see what's popular, it works. If you depend on any one audio fingerprint matchup being accurate, especially accurate enough to use for legal notices, it doesn't make sense.
Music is a semantic object. Saying whether two pieces of music are the same thing depends on stuff that even humans have a hard time figuring out, like how much originality there is in a tribute band's cover.
that's not just reductio ad absurdem, it's actually useful. you should always write the least code possible, and since features mean code, you should have as few features as you can get away with.
I have converted the design document to HTML and done some minor cleanup -- that's here.
Also, there's a tarball of the comple original Nullsoft site, including HTML, source, etc, here. To make your own mirror just unzip the tarball in a public directory.
Agreed. The difference is that Groove has proxying to make sure that offline nodes can still communicate, which makes a huge difference, and about ten million lines more code.
W.A.S.T.E. gets big points for style and simplicity, but doesn't break any new ground.
Slashdot Mirror
Yup, you are missing something. Check out CC Mixter's tagline -- "mixversation".
Remixing is about bouncing ideas back and forth, getting in a conversational flow, a jam approach that isn't done in real-time. It's like a mailing list or a forum. Here and there you see really cool things, the rest of the time it's just people talking.
The idea of being a struggling musician is a bogus way of thinking. The practice of music is not about hustling to become a star. It's about making music constantly, and once in a while making some money on it.
By throwing nodes at the problem. The virtue of P2P is that it scales up like nothing else. The disadvantage is that you need a lot of nodes. Given enough nodes, though, there is nothing with even remotely comparable uptime.
My parents are huge people who bitterly resented the the fact that they were going to have to take me to Star Wars.
N.B.: I don't believe that use of unauthorized music is stealing. It's actually important to split that hair.
I welcome your pea-sized mind to this earth, Brother SnakeStu. Be at peace.
Tip: there's no complicated issue here. The not-very-complicated issue you're referring to (the default copyright on anything copyrightable) is not relevant.
The sense in which you are wrong is that a license is a different thing than copyright. Your statement that a rights holder gets rights automatically is true but has no effect on the licensing situation, which is the comment you were trying to refute. ...relevance of demo scene to OMR, well, uh, SHIT! MY MIND JUST BLEW! ...where's that eyeball... over there on the floor, hm, there it is... *pop* (back in). I dunno, can't see how that issue came into the conversation.
Are you flaming me, Brother SnakeStu?
To make this kind of thing work, you need a lot of candidate songs to find one good song, for whatever definition of good you want. You take a massive number of songs, throw them at the wall, and a few stick. The numbers that OMR had, in the hundreds, weren't even close to the kind of numbers you needed. Webjay has about 50K songs now, and people toss in a few hundred new ones a day.
That's the way it is with collaborative filtering -- you either have a huge amount of input data or the system doesn't work at all.
(About the licensing thing, uh, yeah, well, thanks, but first of all you're wrong and second of all even if you're weren't wrong it wouldn't be relevant. The issue is that OMR couldn't use music from the demo scene because most of that music is deliberately released without any documentation describing the licensing conditions.)
The difference is in the scale of things. No flame intended, but Webjay knows about orders of magnitude more songs than OMR.
Most people in the music industry don't know anything about open licenses. They've barely heard of Linux. Their first reaction when they do hear about open licenses is to recoil.
Then, there's a huge amount of public domain music from the demo scene. But the demo scene culture doesn't use licenses at all, usually -- people just post their music without a license.
Then, there's a lot of music from areas where the GPL has no mindshare, and there's a lot of historical music from before there existed things like software licenses.
The original poster got the point of the thing exactly wrong. This isn't about unauthorized music, it's about sticking to music which is both free and authorized.
Filesharing networks are full of hit songs which are unauthorized. Out on the public/stable web there's a huge amount of unknown stuff which is authorized. The trick with Webjay and other sources mentioned in the article is finding the few great songs in the whole sea of crud. If you can do that, you can have good music which the rights holder doesn't mind you having, though you usually have to give up on name brand musicians.
There's also some kind of Gnu-ish angle that I've never been able to articulate well... It's something like -- if you *choose* to listen to music that isn't from insane powermad label types, you get (more) liberty, and if you choose to listen to non-free music, you give up liberty. That's not quite right because almost none of this music is under a free or open license, though.
FWIW I'm the author of Webjay.org and was quoted in the story.
It's not affiliated with Real in any way. Real playlists do have a bunch of functionality that Winamp and WMP playlists don't, because Real is the only major vendor with half decent (buggy but still usable) SMIL support. As a result playlist authors who depend on SMIL features also tend to have a preference for Real URLs. ...the alternative is proprietary MS video, which is no better.
There's a fair amount more video there, but I agree that it's too hard to find. Keep in mind that this video playlisting thing is user-driven and took me by surprise.
About your other comments on e.g. "it's just a list of mp3s", that's exactly wrong. The client side remix of multiple overlapping media objects is where the fun stuff happens. That depends on SMIL, which depends on Real, so you haven't seen what people are talking about.
(Not exactly true that you have to have Real for SMIL. You could use QuickTime, except that QT's SMIL support is painfully buggy and crash-prone. You could also use proprietary IE extensions.)
The bad buffering is because clients all suck so bad. Within the community of hackers working on this stuff nobody has taken on the really crucial job of making a client that doesn't suck. (VLC obviously doesn't suck, and all the components you'd need are there, but there's no SMIL support.)
Speaking as the author of webjay here:
On a technical level, what's original is that the remixing all happens on the client side. It's a *client side remix*, which is a new thing.
Check out this playlist for a fancier set of techniques, including clipping, multiple audio and video sources at the same time, and a good playlist in general. When you watch that the thing to realize is that the soundtrack is coming from one place, the picture from another, the video from another, and all of that is getting mashed together on *your* machine.
Webjay doesn't alter the original. All it does is pass along a URL with some metadata about where to start and stop. That's about as much of a modification as the height attribute on an img tag.
(Speaking as the author of Webjay)
If pine uses lynx to view html mail, isn't that a violation? Seems like you have to pay off Eolas if you want to do things the Unix way, in lots of small modules that do one thing well.
I just gave Coleman twenty five bucks. You should too.
Contribute to Coleman online: http://colemanforsenate.com/
If you contribute, be sure to call the campaign office and say why, e.g. that this is about his stance on the RIAA. The number is (651) 556-1846. They'll take your comment very seriously when there's money behind it.
If you're a Democrat, like me, just lay down your guard and be a single issue voter for a minute. If getting slashdotted means money for politicians, they'll start taking us seriously.
The issue is how good the re-encoded bits sound. Reputedly they sound pretty good.
Note also that if you use the same encoder as the original, pre-DRM, file, then there shouldn't be a perceptible difference.
Speaking as the author of the oreillynet blog, the news is that an anonymous poster revealed the vulnerability that I had given up on finding. Except that followup gossip about the exploit posted on the blog is that it's not the same one I was trying to find, meaning that there are two, but only one has been publically exposed.
You don't get it. What Relatable (among others) can do is provide a pretty good guess. What it can't do is provide facts. Guesses are useful in aggregate, as statistics, but useless in court.
Audio fingerprinting is not something like a hash function that leads to a deterministic identifier. It is more like a web search engine that finds the best fuzzy match.
If you use audio fingerprint scores in the aggregate, for example to see what's popular, it works. If you depend on any one audio fingerprint matchup being accurate, especially accurate enough to use for legal notices, it doesn't make sense.
Music is a semantic object. Saying whether two pieces of music are the same thing depends on stuff that even humans have a hard time figuring out, like how much originality there is in a tribute band's cover.
The modification of the public key to make it divisible by three was absolutely beautiful. Huge props to the unknown hacker.
that's not just reductio ad absurdem, it's actually useful. you should always write the least code possible, and since features mean code, you should have as few features as you can get away with.
Also, there's a tarball of the comple original Nullsoft site, including HTML, source, etc, here. To make your own mirror just unzip the tarball in a public directory.
Agreed. The difference is that Groove has proxying to make sure that offline nodes can still communicate, which makes a huge difference, and about ten million lines more code.
W.A.S.T.E. gets big points for style and simplicity, but doesn't break any new ground.
This is just plain wrong. The source was never available, leaked or otherwise.
The protocol was reverse engineered, with a little assistance on IRC from deadbeef.