Metadata is a sticky note on a sheet of paper. It may make some statement about what the paper it is stuck to, but it is not written on the paper itself.
Metadata basically means data about data.
The core issue is that if you have a string of bits (say a long line of zeroes), you can't say from that string of bits alone where it came from, how it was created or anything of that sort. You will need metadata/color/sticky-note to tell you this.
Yes you have file formats that can carry the metadata alongside the data (id3 tags in mp3 files for instance). But even if you strip out that metadata the data itself is still there and still usable for whatever you want to do with it.
That still doesn't explain how metadata is color in this context. From the OP:
Those questions are perhaps answerable by "metadata", but metadata suggests to me additional bits attached to the bits in question, and I'd like to emphasize that I'm talking here about something that is not properly captured by bits at all and actually cannot be, ever.
Hrmf, has it really been that long since i read the article. Had completely forgotten that he touched on the subject of metadata. I stand, corrected, i guess...
Though now, after re-reading the author's statement, I'm not sure I agree with all of it. How can anything not be representable with bits? It's just a matter of finding an encoding scheme.
I guess perhaps the author's point is that this coding scheme must be communicated through some side-channel, coloring the bits. The bits themselves could tell you, for instance, which version of a given protocol to use, but couldn't indicate the protocol out of the blue without some other agreement.
And this gets back to the discussion of intent. These bits over here are colored by intent. You can codify that color in bits over there, but those are still different bits, and those bits still need to have color information. It's colors all the way down!
"How can anything not be representable with bits?"
Because copyright law (at least in the U.S.) is bat-$#!+ crazy.
"Copyrights cover copying. If you write a novel that turns out to be word-for-word the same with Gone with the Wind and you can prove you never saw Gone with the Wind, that would be a defense to any accusation of copyright infringement." -- Software patents - Obstacles to software development, Richard Stallman
Metadata is not 'Colour' as the article describes it. Colour is "Who created the bits? Where did they come from? Where are they going? Are they copies of other bits?"
Let's use an example: an mp3 of The Eagles' song Journey of the Sorcerer. The Eagles created the song/are 'where the bits came from' (according to copyright law). This can be encoded in metadata. But metadata can be changed. The fact that The Eagles made the song cannot be changed, and thus cannot be reliably encoded into bits. [I think that's what the author is saying.]
Metadata basically means data about data.
The core issue is that if you have a string of bits (say a long line of zeroes), you can't say from that string of bits alone where it came from, how it was created or anything of that sort. You will need metadata/color/sticky-note to tell you this.
Yes you have file formats that can carry the metadata alongside the data (id3 tags in mp3 files for instance). But even if you strip out that metadata the data itself is still there and still usable for whatever you want to do with it.