Read up on Kolmogorov complexity.

Knowing that, say, some file contains a sequence of zeros you can run through RLE is information. That part is usually not part of the information content calculation, but without it you can't do compression because the file could just be 100% entropy. Ergo it's useful information.

That was sort of my original point anyway, knowing you don't know something makes it a known unknown instead of an unknown unknown. Or in this case a known known.

Yes. The encoding/encoder tuple is part of the definition of K complexity. Sipser gives a lucid explanation in his textbook and I highly recommend it.