7 ms·

It would be silly if actual human file these claims so the claims are very likely to be filed by some algorithm, which somehow found matching pattern in the noi

It would be silly if actual human file these claims so the claims are very likely to be filed by some algorithm, which somehow found matching pattern in the noise. Maybe: 1) Some copyrighted music has a period of noise that closely matched the segment of noise of the video. 2) There is adversarial pattern in the noise in the video that confused the hunting algorithm.

JoeAltmaier 9y ago
Neural networks are easily fooled: http://www.evolvingai.org/fooling http://www.evolvingai.org/fooling
- njharman 9y ago
  Almost as easily as humans.
stordoff 9y ago
Three of the claims are for other white noise recordings. It's arguably not even an algorithmic mistake, but a result of someone claiming they have rights over a recording of white noise.
- cpncrunch 9y ago
  There's nothing wrong in theory with someone claiming copyright over white noise. If someone else blatantly copies their white noise, they are perfectly within their rights to file a copyright claim. However in this case it does seem to be youtube's content matching algorithm that is finding the match (as far as I can tell, but it's hard to know for sure). My guess is that their matching algorithm just doesn't work well for white noise.
  - njharman 9y ago
    Copyright, in USA, requires a creative component. Facts, randomness, and other stuff lacking any creative expression is not covered by copyright. The bar is ridiculously low as in you can claim copyright on particular formatting or page layout of say names and phone numbers. I could imagine, it's a stretch, a specific "arrangement" of white noise being copyrightable. But in general it is wrong to think you can copyright white noise.
    - cpncrunch 9y ago
      There is always going to be some kind of creative component in actually generating the white noise. And even if you argue that white noise is theoretically uncopyrightable, you have to remember that it is also theoretically impossible to prove that any particular piece of sound is actually truly random.
    - Pulcinella 9y ago
      As you said, I could imagine that maybe, maybe a specific white noise file is copyrightable. This would be similar to how typeface copyrights work. The specific font file for Helvetica Neue is copyrightable, but printing out the letters, tracing them, and digitizing that to make your own font is legal (in America at least).
    - brooklyn_ashey 9y ago
      I'm pretty sure John Cage's works are under copyright, which is interesting since he made such a big show of erasing himself from the work and embracing the random. It never occured to me how obtuse that was. I'm thinking particularly of 4'33". So, I guess the 'him' part of that composition is the fact that someone sits at a piano in particular, rather than say... bagpipes or even pipe organ. I wonder if it would be considered satire or a derrivative work on ukelele choir...I wish he was alive to ask him his thoughts on this.
      - visarga 9y ago
        /offtopic > I'm thinking particularly of 4'33". I read that using Apple's Text to Speech (macOS High Sierra) like: > I'm thinking particularly of 4 feet 33 inches. LOL Apple. You haven't heard of Classical Music yet, have you?
- theoh 9y ago
  What's wrong with copyrighting a particular concrete example of white noise? A particular string of randomly generated bits? The fact that an algorithm just hears generic white noise is the algorithm's problem.
  - abritinthebay 9y ago
    > What's wrong with copyrighting a particular concrete example of white noise? It would be functionally impossible for a start. Copyright does not cover ideas and information themselves, only the form or manner in which they are expressed. Given white noise is a random signal having equal intensity at different frequencies it's basically like saying "what's wrong with copyrighting a particular number?" it is nonsensical.
    - FartyMcFarter 9y ago
      > "what's wrong with copyrighting a particular number?" We do it all the time - a file on a hard drive is basically a number, and files are definitely copyrightable.
    - reificator 9y ago
      When you're dealing with copyrighted files, every single one is just a very large number.
      - kbenson 9y ago
        I'll see your large number and raise you an army of monkeys on typewriters, which if I follow the pseudo-intellectual logic going on in this thread correctly, means I own the copyright of everything that will ever be created in the future. Or maybe we can all agree that just being able to describe, view, feel, smell or listen to something doesn't necessarily make it something that we can copyright or patent. We can walk back from the brink of madness towards sanity, it just takes the first few steps. :)
      - chowells 9y ago
        It's not the number that's copyrighted. It's the content that number represents that's copyrighted.

JoeAltmaier 9y ago

Neural networks are easily fooled: http://www.evolvingai.org/fooling http://www.evolvingai.org/fooling

njharman 9y ago
Almost as easily as humans.

stordoff 9y ago

Three of the claims are for other white noise recordings. It's arguably not even an algorithmic mistake, but a result of someone claiming they have rights over a recording of white noise.

cpncrunch 9y ago
There's nothing wrong in theory with someone claiming copyright over white noise. If someone else blatantly copies their white noise, they are perfectly within their rights to file a copyright claim. However in this case it does seem to be youtube's content matching algorithm that is finding the match (as far as I can tell, but it's hard to know for sure). My guess is that their matching algorithm just doesn't work well for white noise.
- njharman 9y ago
  Copyright, in USA, requires a creative component. Facts, randomness, and other stuff lacking any creative expression is not covered by copyright. The bar is ridiculously low as in you can claim copyright on particular formatting or page layout of say names and phone numbers. I could imagine, it's a stretch, a specific "arrangement" of white noise being copyrightable. But in general it is wrong to think you can copyright white noise.
  - cpncrunch 9y ago
    There is always going to be some kind of creative component in actually generating the white noise. And even if you argue that white noise is theoretically uncopyrightable, you have to remember that it is also theoretically impossible to prove that any particular piece of sound is actually truly random.
  - Pulcinella 9y ago
    As you said, I could imagine that maybe, maybe a specific white noise file is copyrightable. This would be similar to how typeface copyrights work. The specific font file for Helvetica Neue is copyrightable, but printing out the letters, tracing them, and digitizing that to make your own font is legal (in America at least).
  - brooklyn_ashey 9y ago
    I'm pretty sure John Cage's works are under copyright, which is interesting since he made such a big show of erasing himself from the work and embracing the random. It never occured to me how obtuse that was. I'm thinking particularly of 4'33". So, I guess the 'him' part of that composition is the fact that someone sits at a piano in particular, rather than say... bagpipes or even pipe organ. I wonder if it would be considered satire or a derrivative work on ukelele choir...I wish he was alive to ask him his thoughts on this.
    - visarga 9y ago
      /offtopic > I'm thinking particularly of 4'33". I read that using Apple's Text to Speech (macOS High Sierra) like: > I'm thinking particularly of 4 feet 33 inches. LOL Apple. You haven't heard of Classical Music yet, have you?
theoh 9y ago
What's wrong with copyrighting a particular concrete example of white noise? A particular string of randomly generated bits? The fact that an algorithm just hears generic white noise is the algorithm's problem.
- abritinthebay 9y ago
  > What's wrong with copyrighting a particular concrete example of white noise? It would be functionally impossible for a start. Copyright does not cover ideas and information themselves, only the form or manner in which they are expressed. Given white noise is a random signal having equal intensity at different frequencies it's basically like saying "what's wrong with copyrighting a particular number?" it is nonsensical.
  - FartyMcFarter 9y ago
    > "what's wrong with copyrighting a particular number?" We do it all the time - a file on a hard drive is basically a number, and files are definitely copyrightable.
  - reificator 9y ago
    When you're dealing with copyrighted files, every single one is just a very large number.
    - kbenson 9y ago
      I'll see your large number and raise you an army of monkeys on typewriters, which if I follow the pseudo-intellectual logic going on in this thread correctly, means I own the copyright of everything that will ever be created in the future. Or maybe we can all agree that just being able to describe, view, feel, smell or listen to something doesn't necessarily make it something that we can copyright or patent. We can walk back from the brink of madness towards sanity, it just takes the first few steps. :)
    - chowells 9y ago
      It's not the number that's copyrighted. It's the content that number represents that's copyrighted.