What Colour are your bits? (2004)

If curious see also

2014 https://news.ycombinator.com/item?id=8728532

2013 https://news.ycombinator.com/item?id=6231992

2009 https://news.ycombinator.com/item?id=506986

This is actually very helpful to mentally understand why people who are not thinking with regards to the logic that software works at don’t see eye to eye on these issues. I would have never imagined something like nuance on copyright would somehow weave back around into opinion about obscenity law, and yet this is about as elegant as any explanation of it.

This is my first time reading this, but I suspect it has become a new part of my mental modeling of the world for years to come.

I think it would be better to think in terms of information theory, rather than the hypothetical color of bits. If I have a legally obtained copy of a copyrighted song (that is not permissively licensed), then I am restricted in my right to share copies. Certainly I can create a one-time pad and communicate that to someone, and I can XOR the bits of the song recording with that one-time pad, and the author would argue that I now have colorless bits. But I send it to you along with instructions for how to decrypt it, I'm using a mechanism to communicate a perfect copy. It isn't that the ciphertext is now "colored", it's that the net effect of my mechanism is that I've communicated the original song, with perfect fidelity. That's what matters: the net effect of the system as a whole. I will have created a communication channel (perfectly legal), and used that channel to share a copyrighted work without the creator's permission (not legal, depending on the details).

Could we add [2004] please? This is a great article, always glad to see it pop up again!

Colour does exist. It's the Second Law of Thermodynamics: information cannot be created, only lost, and creating a faithful copy of document A requires you to discard an equivalent amount of information (usually in the form of “there's usable energy over here”, in a very lossy process, but physics doesn't require this) – which you can then recover by comparing your copy A with the original document A, and discarding one.

You're not going to get the Complete Works of Shakespeare by any means other than by copying an existing copy of the Complete Works of Shakespeare; the odds are astronomical. You might get a description of calculus by other means (expending the effort yourself to produce it), but it's not very likely you'd come up with it before somebody else made such a discovery public, unless:

• it's recently become obvious that such a thing would be useful; and

• it's easy enough to discover that you could figure it out in a year or two; and

• enough people were looking in that direction at the time (or one of them was a secretive sort of person).

That's not to say that authorial monopolies are necessarily a good thing. But they are a meaningful concept.

The blog post writes:

> Suppose you publish an article that happens to contain a sentence identical to one from this article, like "The law sees Colour." That's just four words, all of them common, and it might well occur by random chance. Maybe you were thinking about similar ideas to mine and happened to put the words together in a similar way. If so, fine. But maybe you wrote "your" article by cutting and pasting from "mine" - in that case, the words have the Colour that obligates you to follow quotation procedures and worry about "derivative work" status under copyright law and so on.

There was a real court case in 2012 which I think is interesting because it's very similar to this example. A photographer was accused of "copying" the concept of taking a photo of a red bus in front of a grey Houses of Parliament. He defended himself by saying that that those ideas are very common and should not be copyrightable---but failed:

https://youzicha.tumblr.com/post/162846191544/what-colour-ar...

I just read this about 5 years after it was published.

Working with computers the whole life and graduating in Law, I always had the problem of not understanding how my colleagues could think that the bits from one email where different from the bits of something else when you were doing a dd from one disk, or how reading one disk was "breaking" some correspondence secret....

This is an elaboration of the idea that where data comes from is just as important as what the data happens to be. In expert systems this is related to the logical justification of a truth statement, as the statement is still true or false, but there is a Pedigree or Provenance for the data provided as metadata about the data stored.

There is the thing itself and there is the history and context of the thing, and they are both important. We try to keep them together, but the connection is often fragile.

One example: a product and its price. The price roughly summarizes information about how it was made and how valuable people think it is, which are not attributes of the thing itself and often can’t be deduced from it. In some cases we physically attach a label, the price tag, to keep track. In other cases we attach a lot more info.

Another example: a photo and information about when and where it was taken. For a photo to serve as evidence we need a reliable history. In court, this is the chain of custody. Too often on the Internet, we pretend that having the photo is proof enough, but without knowing its history, it could be a fake.

This is a great metaphor because you will often see people with a technical background arguing copyright law is if it was purely about the bits. Because for us, the bits are the only real thing. But the law does not see it like that.

I've wondered how long it will be before input events are traceable to specific devices (e.g., you type 'A' on your keyboard, registered to you, and get a scancode . . . plus a few hundred bits timestamped and signed by the device before anyone can actually treat it as an 'A' keystroke). "Secure Unicode," anyone?

I figure this kind of dystopian mechanism for input provenance is at least 100 years out. Please don't anyone prove me wrong.

Medium grey and opaque, thanks to cryptography.

It'll be time to update to 2020!

this is almost a description of a type system

This essay is written to create the impression that it's imparting something profound, but it's really just identifying the existence of side channels.

We encode bits in all kinds of things. You can store some bits on a flash drive. Then you can write some words on the outside of the flash drive. They're both just bits.

Whether you own a flash drive with some bits on it which is in a safety deposit box at a bank may depend on the bits in the bank's computer and not the bits on the flash drive, but it's still bits that it depends on.

The examples it uses aren't accurate:

> Maybe you were thinking about similar ideas to mine and happened to put the words together in a similar way. If so, fine. But maybe you wrote "your" article by cutting and pasting from "mine" - in that case, the words have the Colour that obligates you to follow quotation procedures and worry about "derivative work" status under copyright law and so on.

As it turns out copyright law doesn't really care about this, even if people might expect it to, because the law is too pragmatic for that. Proving that you came up with some particular phrasing is hard. Disproving it is hard too. So in practice the courts don't look at whether you actually copied the bits. They look at whether you could have copied them ("access"), which they then go on to assume is the case for anything widely disseminated (since establishing that would be hard too), and then whether the bits are similar ("substantial similarity"):

https://www.theiplawblog.com/2007/02/articles/copyright-law/...

Whether you actually copied the bits doesn't come into consideration, apparently.

Because the courts can only take into consideration the information they have available to them. Which is all bits, because all information is bits.

> You take a file to which someone claims copyright, mix it up with a public file, and then the result, which is mixed-up garbage supposedly containing no information, is supposedly free of copyright claims even though someone else can later undo the mixing operation and produce a copy of the copyright-encumbered file you started with.

There is still no Colour here, and the essay is missing a rather decent practical attack on the "Colour theory" version of the copyright system.

Suppose Alice publishes R1 and Bob publishes R2. R1 is Alice's message xor Alice's one time pad. R2 is Bob's message xor Bob's one time pad. The one time pads are "random".

Then it's discovered that R1 xor R2 generates a third party copyrighted work. Which is statistically impossible unless either Alice or Bob (but not necessarily both) chose their one-time pad specifically in order to cause this.

According to "Colour theory" the one who chose their one-time pad specifically in order to cause this has given their ciphertext the "Colour" of the copyrighted work. But that's a real problem in practice when there is no way to tell which one it was. The other person may not even be in on it. So which one do you haul into court when you don't know that? Which one do you take down?

In real life if something like that becomes popular what happens is not that they figure out who it really was that created the derivative work, it's that they come up with some kind of disgusting hack like the DMCA takedown process which imposes no practical consequences on fraudulent takedowns, and hope that the innocent victims of the collateral damage don't have enough political clout to do anything about it.

It seems like the same fallacy as the model of the law that we teach to high school students. Computer scientists understand that it's wrong. Lawyers understand that it's wrong. But certain people benefit from pretending that it isn't in front of the general population because the misleading abstraction is prettier than what actually happens under the hood, and a better understanding of the latter would make people upset.