8 ms·

The main problem with "illegal numbers" seems to be that for any illegal number a, which represents someone's intellectual property or trade secret A, I can eas

by rolux 14y ago

The main problem with "illegal numbers" seems to be that for any illegal number a, which represents someone's intellectual property or trade secret A, I can easily create two original pieces of intellectual property B and C so that b + c = a.

001sky 14y ago
This is a good point, as it the derivative notion: are my thoughts about illegal numbers, themselves, illegal. Evidentiary issues aside. etc.
wmf 14y ago
This is covered in "color", already posted in this thread.
Evbn 14y ago
Give us an actual example of b and c that are created without knowledge of a, and you may have case. But information theory says it is unlikely, if a is deeply original and not derivative of something like b and c to begin with.
- aes256 14y ago
  I may not be on the same wavelength as rolux here, but compression would be one example. Create B (a series of compressed files) and C (a decompressor), both unique creations in their own right, that when combined create the original work of another person, A. Now I'm no expert on compression, but there are surely a vast, if not infinite, number of possible pairs B and C that could combine to form the original copyrighted work. Thus, with copyright law in its current state, we are not simply granting copyright holders the exclusive rights to a particular sequencing of 1s and 0s, but also rights over any method of creating that sequence. If there are an infinite number of methods of creating that sequence — that is to say, any B, if combined with a suitable C could form A — then aren't we in effect granting rights over everything to the copyright holder? Where do we draw the line? The 1s and 0s that make up this post could, with suitable decompression, form an exact copy of a hit blockbuster, but neither this post nor the decompressor would resemble the blockbuster on their own. Edit: Interesting idea time. If I uploaded a series of files alleging they are encrypted copies of blockbuster films, but resolved not to release the encryption keys to any of the files for 12 months, would the copyright holders have the right to have the encrypted files taken down in the meantime? They can't actually prove that the files are infringing without the encryption key. Is the mere suggestion that a file may potentially, with some manipulation (i.e. decryption with the appropriate key), resemble a copyrighted work, sufficient to have it taken down?
  - wmf 14y ago
    Again, that can only work if either B or C is a derivative of A.
    - aes256 14y ago
      Apparently what I was describing here has already been put into practice in the shape of Monolith [1] In the case I describe, as in the case of 'munging' using Monolith, neither B nor C bears any resemblance to A except when the two are combined. In my view, at least, you cannot say either is a derivative of A. To do so would bind you to declare everything a derivative of A, because any B when combined with an appropriate C can form A. If I take your post here as B, you will likely deny it is a derivative of any copyrighted work, but if I make the text of the post the encryption key to another file (C) which, when decrypted, becomes copyrighted work A, is your post itself derivative of a copyrighted work? What makes your post non-derivative and the encrypted file I create derivative? They are both nothing in themselves, and yet a copyrighted work when combined with the other. [1] http://monolith.sourceforge.net/ http://monolith.sourceforge.net/ Edit: Reading through the article on the color of bits it seems this exact argument prompted the article in the first place. I guess I should finish reading this!
      - wmf 14y ago
        Let me put it a different way. If A = decompress(B), then necessarily B = compress(A), so B is obviously a derivative of A. Introducing xor does not change anything; one of the parts must be a derivative of the original.
        aes256 14y ago
        Okay, let me propose an alternative procedure. I set a series of random number generators going, and with each set of results, I apply randomly generated XOR to create a new sequence of numbers. I perform this process over and over. Eventually, it produces (give or take a few bits) a copy of an MP3 file of a copyrighted work. Now, once we've eliminated any procedure of creating B and C that includes A, would you still say one of B or C are derivative of A? Should my random number generator be banned? Perhaps more importantly, do I acquire copyright to all the files it creates? I could quite easily create every possible variation of an MP3 file of a given length. Does that mean any musician who, using a different procedure, produces one of these files is infringing on my copyright?
- lmm 14y ago
  I've read an argument that the law has considered a five(?)-note sequence enough to identify one song as a derivative of another - and given reasonable estimates for the "space" of possible 3-minute tunes, and the number of (copyrighted) songs currently in existence, there's a significant chance that a "random" new song will be legally considered a derivative of some existing song. Of course, here we come back to the "color" of bits; one is unlikely to be prosecuted for "copying" a song one has never heard. Though cases like My Sweet Lord are enough to give one pause.
- randomdata 14y ago
  If (b) and (c) can be considered derivative works, then should a low-bit hash of (a) not also be considered a derivative? If a given hashing algorithm says the hash of Metallica's latest hit is 42, have I just broken the law by posting this?
  - Dylan16807 14y ago
    It's not just a yes/no on derivation, the amount of information present in the derivative data matters. If you hashed every 32 bits individually then you would be infringing, but one hash for the whole thing is fine.
a3_nm 14y ago
As a general rule, using secret sharing protocols, for any choice of n and k, you can split any illegal piece of information a into n pieces such that k of these pieces together can reconstitute a but <k pieces will not give any information about a. It is hard to tell to what extent an individual piece is illegal or not. Related: http://www.madore.org/~david/misc/freespeech.html http://www.madore.org/~david/misc/freespeech.html (does something similar using XOR, and discusses interesting uses of that kind of system).

by rolux 14y ago

001sky 14y ago

This is a good point, as it the derivative notion: are my thoughts about illegal numbers, themselves, illegal. Evidentiary issues aside. etc.

wmf 14y ago

This is covered in "color", already posted in this thread.

Evbn 14y ago

Give us an actual example of b and c that are created without knowledge of a, and you may have case. But information theory says it is unlikely, if a is deeply original and not derivative of something like b and c to begin with.

aes256 14y ago
I may not be on the same wavelength as rolux here, but compression would be one example. Create B (a series of compressed files) and C (a decompressor), both unique creations in their own right, that when combined create the original work of another person, A. Now I'm no expert on compression, but there are surely a vast, if not infinite, number of possible pairs B and C that could combine to form the original copyrighted work. Thus, with copyright law in its current state, we are not simply granting copyright holders the exclusive rights to a particular sequencing of 1s and 0s, but also rights over any method of creating that sequence. If there are an infinite number of methods of creating that sequence — that is to say, any B, if combined with a suitable C could form A — then aren't we in effect granting rights over everything to the copyright holder? Where do we draw the line? The 1s and 0s that make up this post could, with suitable decompression, form an exact copy of a hit blockbuster, but neither this post nor the decompressor would resemble the blockbuster on their own. Edit: Interesting idea time. If I uploaded a series of files alleging they are encrypted copies of blockbuster films, but resolved not to release the encryption keys to any of the files for 12 months, would the copyright holders have the right to have the encrypted files taken down in the meantime? They can't actually prove that the files are infringing without the encryption key. Is the mere suggestion that a file may potentially, with some manipulation (i.e. decryption with the appropriate key), resemble a copyrighted work, sufficient to have it taken down?
- wmf 14y ago
  Again, that can only work if either B or C is a derivative of A.
  - aes256 14y ago
    Apparently what I was describing here has already been put into practice in the shape of Monolith [1] In the case I describe, as in the case of 'munging' using Monolith, neither B nor C bears any resemblance to A except when the two are combined. In my view, at least, you cannot say either is a derivative of A. To do so would bind you to declare everything a derivative of A, because any B when combined with an appropriate C can form A. If I take your post here as B, you will likely deny it is a derivative of any copyrighted work, but if I make the text of the post the encryption key to another file (C) which, when decrypted, becomes copyrighted work A, is your post itself derivative of a copyrighted work? What makes your post non-derivative and the encrypted file I create derivative? They are both nothing in themselves, and yet a copyrighted work when combined with the other. [1] http://monolith.sourceforge.net/ http://monolith.sourceforge.net/ Edit: Reading through the article on the color of bits it seems this exact argument prompted the article in the first place. I guess I should finish reading this!
    - wmf 14y ago
      Let me put it a different way. If A = decompress(B), then necessarily B = compress(A), so B is obviously a derivative of A. Introducing xor does not change anything; one of the parts must be a derivative of the original.
      - aes256 14y ago
        Okay, let me propose an alternative procedure. I set a series of random number generators going, and with each set of results, I apply randomly generated XOR to create a new sequence of numbers. I perform this process over and over. Eventually, it produces (give or take a few bits) a copy of an MP3 file of a copyrighted work. Now, once we've eliminated any procedure of creating B and C that includes A, would you still say one of B or C are derivative of A? Should my random number generator be banned? Perhaps more importantly, do I acquire copyright to all the files it creates? I could quite easily create every possible variation of an MP3 file of a given length. Does that mean any musician who, using a different procedure, produces one of these files is infringing on my copyright?
lmm 14y ago
I've read an argument that the law has considered a five(?)-note sequence enough to identify one song as a derivative of another - and given reasonable estimates for the "space" of possible 3-minute tunes, and the number of (copyrighted) songs currently in existence, there's a significant chance that a "random" new song will be legally considered a derivative of some existing song. Of course, here we come back to the "color" of bits; one is unlikely to be prosecuted for "copying" a song one has never heard. Though cases like My Sweet Lord are enough to give one pause.
randomdata 14y ago
If (b) and (c) can be considered derivative works, then should a low-bit hash of (a) not also be considered a derivative? If a given hashing algorithm says the hash of Metallica's latest hit is 42, have I just broken the law by posting this?
- Dylan16807 14y ago
  It's not just a yes/no on derivation, the amount of information present in the derivative data matters. If you hashed every 32 bits individually then you would be infringing, but one hash for the whole thing is fine.

a3_nm 14y ago

As a general rule, using secret sharing protocols, for any choice of n and k, you can split any illegal piece of information a into n pieces such that k of these pieces together can reconstitute a but <k pieces will not give any information about a. It is hard to tell to what extent an individual piece is illegal or not. Related: http://www.madore.org/~david/misc/freespeech.html http://www.madore.org/~david/misc/freespeech.html (does something similar using XOR, and discusses interesting uses of that kind of system).