"Colour" is just history. Both bits and physical things can have history that isn't recorded in the thing itself, but is relevant to the law. For example, the law cares if you're driving a stolen car, even though the car's atoms don't carry XML tags saying "stolen". Enforcing rules based on history is hard, both in the physical and digital world, but it's not impossible. It might even be easier in the digital world, due to ubiquitous surveillance and robust content ID algorithms. Pointing to computer science is just a deflection, like a theoretical physicist explaining that protons are indistinguishable when asked about tracking stolen cars. In practice, history can be tracked. The only question is whether we want to.
1. History in the real world is very much traceable. That's what a paper (celulose or digital) trail is. The history may not be physically attached to the object but the interested parties should have the appropriate property documents. And when there are contradictory claims, documents trump spoken agreements.
2. In the physical world no to objects (above atomical level if you want to be pedantic) are the same and there is a limited supply. In the digital world, a copy is absolutely identical and indistinguishable from its original and from the potential infinity of other copies.
I don't know what you mean by "in practice", but if everyone in a network is always either uploading encrypted bits or random bits, then the problem of tracking the history of that network reduces, in practice, to the problem of distinguishing between the two.
It's not like it's unheard of for people to get arrested for online activity. And from what I understand, those often don't rely on network sniffing.
Because somewhere on your abstraction stack, you'll eventually get to things that are human-meaningful information. Things like facebook or backpage posts, credit card transactions, etc. Things that are in various ways tied to identifiable entities that exist in the physical world.
"Colour" overcomplicates what could be a simple issue. There are bits that describe how to turn it from random data into copyrighted material (eg sound, visuals). This is easy to see. Imagine you took some copyrighted material and encrypted it. Then you posted that online. Without the public key, the file would be useless, and you wouldn't be violating copyright. Now imagine you posted the decryption key. Now anyone can play the file, and you are infringing copyright.
This can be easily applied to the monolith case. If you take a chunk of data out of /dev/random, then this of course is not copyrighted. If you xor it with some copyrighted work, the result is also not copyrighted. If you post either of those files online, you haven't broken any laws. If you post both online, you still haven't broken a law. If you tell everyone to go xor those two files together, then you have suddenly infringed the creators copyright, by allowing others unpermitted access to copyrighted works.
Copyright law differs from place to place, but I think what you have written is incorrect. There is a distinction between "infringement" and "making available". You can infringe copyright without making it available. In this case, making a copy is infringement, even if you don't make it available. Even loading a file into a computer can be an infringement, and so you need a license to do so. The format is irrelevant.
If you take a chunk of data out of /dev/random, it is not subject to copyright because it is not a creative work. Mathematically transforming a creative work does not remove the copyright. Laws are written to be loose enough to deal with corner cases like this.
Out of interest, I once saw a modern art piece that was a sculpture of an ouroboros made of pornographic magazines. Clearly the magazines were still under copyright. So is the sculpture a derived work? I think this is what is known as a transformative use. My understanding is that it is still an infringement, but that it is covered under fair use. (I have no idea how this is covered in countries without fair use!)
I suppose you could encrypt a recording and publicly destroy the key. You could then actively trade the encrypted file. I suspect this would also be a transformative use as a kind of performance art. But as in all fair use, it is still an infringement.
When you xor data with truly random data, the result is just as truly random as the original random data. Consider this: considering that xor is its own inverse, how would a court distinguish the original and the xored random data?
Perhaps you're right, and the latter is legally considered copyrighted. However, on a practical level, this is completely unenforcable; so long as you never state the way the data came about, there is no way to acertain the connection (http://en.wikipedia.org/wiki/one-time_pad).
This is what the lawyer of the article believes: the xored data is copyrighted. He explains it by introducing a metaphysical "colour" associated with the data. Given that litigation can only occur when you have stated that the data is a transformation of a copyrighted work, it seems much simpler to ascribe the litigation to your statement, rather than to "colour".
I am under the impression that "copying", as far as copyright goes, really refers to distributing. You are legally entitled to make a single backup of copyrighted work you come into possession of. If this backup was stored in the cloud, I imagine no one would mind, unless you handed your password out to just anyone. This is why I believe that distributing an encrypted copy is safe - without the password, it doesn't constitute distributing the original.
As you said, copyright is confusing, so I may be wrong on the previous count. However, the fair use claim is much easier than you make it out to be. One clause of fair use is that the work doesn't compete with the original. An encrypted file, being useless for consuming the original, surely doesn't compete with it.
Just keep in mind that "fair use" is not a defence. If you invoke fair use, then you have admitted infringement. If you are awarded fair use, it's just that you are not liable for a penalty.
It's kind of pedantic, but that's the way the law works. The problem comes with the corner cases. If you use a non-pedantic view of the law and think, "Well, it's just common sense that <insert common sense here>", you may very well come up with things that are very, very incorrect (and possibly very damaging to you).
The point is that it is impossible for a computer to reliably determine if your access to a copyrighted material is lawful or infringing.
You can try to fake it and pretend the computer has a way of knowing but you will unavoidably err either on the side of permitting infringing access or on the side of restricting lawful access or both.
DRM errs on both sides by at the same time being ineffective at preventing unlawful duplication (it can bypassed and if it can't, then blocking the analog hole is physically impossible) and over-restricting lawful use (fair-use, different user hardware, etc.). Streaming services are the same.
There is no way for software to encode the multitude of ever changing details of the abomination that is copyright law.
The "Color" analogy is apt because, copyright law is to a computer like color is to a colorblind person. S/he can pretend to recognize colors based on context, brightness, contrast, etc. but s/he can not actually see color. (Yes, I know there are various degrees of color blindness. For the sake of the argument take an extreme case of colorblindness.)
And to be fair, the law is so murky it is similarly hard for humans too.
